Northfield Webinar Directory
Please Contact Northfield Sales or Support if you would like to view the full presentation recordings for any of these webinars.
The slides from each of these presentations are available for download on our website, www.northinfo.com/research.php, under the “Webinar Proceedings Archives” category.
|
Americas Sales: 617.208.2050 Support: 617.208.2080 |
EMEA +44 (0) 17-2244-RISK (7475) |
Asia +61 (2) 9238-4284 |
US Sovereign Debt Default Risk in a Higher Interest Rate Environment
Presenter: Emilian Belev
April 30, 2024
Abstract
A focus of attention of economists and markets increasingly so over the last year has been the simultaneous upward trajectory of the level of US Treasury debt since the pandemic and the sharp increase of treasury interest rates due to tighter monetary policy. The higher debt balance at higher rates means higher periodic interest rate costs. In the meantime, the federal budget seems to be in routine need of external financing which precludes retirement of debt and perpetuates its growth. Interest costs crowd out increasingly higher amounts of non-discretionary expenditure which is then financed with more debt.
In 2013, Northfield claimed the Professional Risk Management International Association (PRMIA's) award for innovation with its model of sovereign credit risk. The paper which outlined the model later became part of a book - World Scientific Reference on Contingent Claims Analysis in Corporate Finance curated by Michel Crouhy, Dan Galai, and Zvi Wiener (2019).
Since then, we have developed new powerful and robust tools that leverage the same principles behind the model to new levels of capability. Using simulation technology that captures an extremely high number of possible outcomes years ahead - a technology not available anywhere but through Northfield - we estimate the probability and severity of a shortfall of government revenues. This is done with respect to both net debt service and total outstanding debt, given the expectations of future debt growth, economic growth, and different tax rate scenarios. We also estimate the value of the sovereign "default option" through time both with and without the potential for accommodative monetary response to infer its impact and the potential trajectory of the future cost of borrowing.
Optimization of Taxable Institutional Portfolios
Presenter: Dan diBartolomeo
April 16, 2024
Abstract
While most asset managers think of taxable portfolios as being part of a “private wealth” business, the current value of US taxable asset owner portfolios (e.g. insurance companies) is about $13 Trillion. Globally, we estimate the value of insurance company AUM at around $36 Trillion. There are also some other types of institutional investors with material assets such as special trust funds set aside for diverse activities (e.g. dismantling of nuclear power plants, payouts to claimants in class action lawsuits). Commercial banks also hold a wide range of securities, although what is allowed on a bank balance sheet varies widely by country. While “tax aware” investing has long been a part of these activities, there are far more overlapping issues such as regulatory requirements and the need for constant liquidity.
In this presentation, we will illustrate how neither “tax dumb” asset management, nor simple techniques like “tax loss harvesting” are sufficient in this context and propose basic methods that will allow more efficient management of taxable institutional portfolios across a range of asset classes.
Cointegration of Sector Returns
Presenter: Dan diBartolomeo
March 28, 2024
Abstract
In this presentation we will explore investment strategies that are based on the statistical concept of cointegration rather than the more familiar concept of correlation.
Cointegration methods are meant to be applied to time series that do not have the statistical property of “stationarity” which is an assumption of many statistical measures such as traditional correlations. For example, it is typical that period by period returns to investment factors are stationary but cumulative returns to those same factors are not stationary. Strategies that use cointegration methods can operate directly on cumulative returns which is often of greater practical interest to long-term investors than the distribution of periodic outcomes.
We have previously discussed cointegration of country level equity index returns and presented evidence that “active” returns can be obtained from passive management methods.
We show that similar positive outcomes can be expected when our methods are applied to sector (or industry) returns within countries.
Why Most Tax Optimizations are Clearly Sub-Optimal
Presenter: Dan diBartolomeo
March 14, 2024
Abstract
Since pioneering tax optimization in 1995, we have seen many efforts that streamline tax sensitive portfolio strategies (e.g. loss harvesting) gain popularity. While that streamlining has made it easier for managers to deliver such strategies, the simplifying assumptions that facilitate such processes do not generally reflect the economic reality of how taxes interact with return and risk of security portfolios.
In this presentation, we will describe how these assumptions do a material disservice to investors who are provided with sub-optimal portfolios.
For example, it is a common industry practice to undertake harvesting tax losses under the assumption that harvested losses always have economic value, even though in the real world the economic value of capital losses is conditional on the investor having capital gains to offset. While excess losses can be carried forward to future years, the presumption of “future gains” also implies a lesser present value of the losses and being subject to limitations on tax-loss carryforwards. Such unrealistic assumptions are sometimes even carried over to distorted reporting of “after-tax returns” with exaggerated “tax alpha.” Similarly, the proper economic tradeoffs between avoiding taxes and risk while maximizing returns involve amortizing the incremental (higher) taxes on short-term gains over a different horizon than the basic rate of taxation imbedded in the lower long-term rates. Even in the presence of the expectation of capital gains in future tax years, the present value argument can also impact the choice of whether to aggregate short-term and long-term capital gains/losses for the purposes of the investor objective function.
A final source of sub-optimality is the economically improper handling of the wash sale rule.
Investment Performance Evaluation (Real and Simulated) Using the Effective Information Coefficient
Presenter: Dan diBartolomeo
February 27, 2024
Abstract
Whether evaluating live or simulated performance, one of the major difficulties in performance evaluation is determining the statistical significance of the results. All too often, practitioners simply ignore this issue and present results whether good or bad, that could have arisen from luck or skill. To make performance analysis useful for improvement of the investment process, understanding how to assess the significance of the results is of critical importance. Often systems simply revert to analyzing performance daily as opposed to weekly or monthly under the assumption that “more observations must be more significant.” For a variety of statistical reasons presented in diBartolomeo (Journal of Performance Measurement, Spring 2003) that assumption is false. We cannot manufacture statistical significance merely by shorting the observation intervals.
A much better way to obtain results with statistical significance is to evaluate the contribution individual positions within the portfolio make to risk adjusted performance metrics. Since the number of active weights in an equity portfolio is typically in the hundreds, sample sizes are much larger compared to the traditional evaluation of portfolio level returns. In diBartolomeo (Journal of Performance Measurement, Fall 2008), the concept of the Effective Information Coefficient was established. The EIC combines three ideas familiar to equity quant managers: information coefficients, breadth, and transfer coefficients, combining them into a unified metric. Within firms, the EIC can be used to evaluate live or simulated performance. Uniquely, the EIC is structured to also be used by outside observers (e.g. clients, consultants) without requiring information known only to managers (e.g. alpha forecasts).
The EIC process has been implemented in the Northfield’s attribution application and will be deployed to the Nexus online platform in the months ahead.
From Measurement to Mastery: Elevating Forecasting Performance in Asset Management
Presenter: Guest Speaker Dr. Armin Grueneich
February 13, 2024
Abstract
Dr. Armin Grueneich will talk about what he learned in the past dozen years about measuring the quality of the forecasts of equity analysts. Simply put, when a stock is at $100 and an analyst predicts it will be at $150 in a year and it ends up at $120, was that a good call? What if the market is down 20% over that period?
The answers to these questions, of course, depend on what other forecasts were available at the time, how success is measured, and for what purpose. Armin's presentation will focus on measurements that can be used in a systematic improvement process. His talk will concentrate on fundamental and quantamental investment processes and how measuring forecast quality can guide a systematic improvement effort. But the ideas that he will present are easily transferable to quantitative strategies or even manager selection.
The presentation will have a nineties vibe.
The Efficient Price of Illiquidity and the Option to Wait
Presenter: Emilian Belev
January 30, 2024
Abstract
The illiquidity premium has long occupied the attention of researchers and investors in liquid and illiquid assets. Estimating the price investors require for foregoing the ability to have sufficient liquid resources in future periods is key to determine the optimal tradeoff between holding liquid and illiquid assets. Previous work in this area by Golts, Kritzman, diBartolomeo and others suggested that the value of illiquidity can be thought of being equal in size and reverse in sign relative to the value of an option that allows the investor to put on hold temporarily the need to expend liquid resources at some future point in time.
This work is consistent with this view and its new contribution is twofold. First, it presents an optimization framework which helps to explicitly determine the price of illiquidity. It does so by finding the expected return premium that will make the investor achieve the same relative utility from an illiquid asset compared to a liquid asset. Secondly, in doing so, it does not impose any assumptions present in familiar option pricing models and does not assume the existence of an option market of sufficient depth.
Approximating Multiperiod Portfolio Optimization
Presenter: Dan diBartolomeo
January 16, 2024
Abstract
The real-world nature of investment portfolio management is that it is a sequence of events and decisions. On the other hand, the traditional way of thinking about portfolio construction is the Markowitz optimization process in which the future is presumed to be a single period of unknown length. If and only if transaction costs and taxes equal zero are these two conceptions analytically equivalent. If there are no frictional costs for changing a portfolio, we can always define the present moment as the start of a new single period whenever conditions change. When transaction costs (and taxes) are non-zero, an investor’s current optimal portfolio is always a blend of what they would hold if they were starting their portfolio anew (i.e. from cash) and their preexisting legacy portfolio. While most optimization processes take transaction costs into account in some fashion, the usual methods are insufficient to represent the true process as an ongoing sequence of future actions, leading to biased estimation what is the true optimal portfolio.
In this presentation, we will illustrate how to address this issue using a concept we call “probability of realization of incremental utility” as implemented in the Northfield Optimizer. Correct usage of this functionality improves portfolio performance by reducing the costs associated with unnecessary portfolio turnover.
We will also discuss how this concept of maintaining optimality over multiple periods can be incorporated in performance attribution.
Performance Attribution When Derivatives or Leverage Are Present
Presenter: Dan diBartolomeo
December 28, 2023
Abstract
When the SEC (US market regulator) enacted Rule 18f-4, it required that mutual funds and ETFs that have any form of leverage in their portfolio, either through use of derivatives or short positions do an estimation of the Value-at-Risk (VaR) metric to monitor potential losses for investors. It is often incorrectly asserted that 18f-4 risk estimation must be done on a “historical simulation” basis, although VaR estimation based on other methods such as factor models or Monte Carlo simulation are permissible. The rule further requires that funds track the correctness of their VaR estimates and make a report to regulators if risk is poorly estimated for a material period.
However, to be useful to investors in terms of improving future portfolio performance, the presence of derivatives or short positions must be appropriately incorporated into performance attribution processes, making their impact on return and risk over time fully transparent.
In this presentation, we will illustrate how the presence of short positions, margin leverage and derivatives should be incorporated into performance attribution and related reporting. Using our internal PWER fund rating system, we will provide an empirical example of ten bonds funds that have made increasing use of Treasury bond futures in recent years.
An Asymmetric Representation of Security Alpha for Active Management
Presenter: Dan diBartolomeo
December 6, 2023
Abstract
The traditional way of describing “alpha” is to assume a change in the expected return mean relative to some neutral expectation (i.e. benchmark or CAPM). It is routinely assumed that the distribution of outcomes around the revised expectation is normal.
In this presentation, we will provide an alternative way to describe security level alpha where the expectation is equal to the benchmark, but the distribution of outcomes has positive or negative skew. In effect, we define alpha by a greater likelihood of the outcome being realized in the upper or lower tail of the ex-ante distribution. This distinction is critical in optimization. In this representation, a positive alpha implies that a security holding also contributes less to risk, and a negative alpha position contributes more to risk.
We will show that this representation is more consistent with studies showing that “high conviction” (i.e. high active share) managers generally produce better outcomes.
We will also show that this representation is more consistent with both practice and empirical outcomes in high risk investing areas such as venture capital, pharmaceutical research, and movie financing.
How Many Bonds Make A Diversified Portfolio? Fixing DTS
Presenter: Dan diBartolomeo
November 28, 2023
Abstract
Many popular models of fixed income portfolio risk utilize a method called “Duration Times Spread” (DTS) as a summary measure of the credit risk embedded in a portfolio. DTS methods implicitly assume that credit risks in portfolios are so well diversified that “downside risk” and “upside potential” are equal (a consequence of the Central Limit Theorem). It should be obvious that a single bond will fall further in value due to a credit default than that same bond will rise in value because it has become riskless from a credit perspective. As such, a middle point must exist between the two ends of diversification spectrum at which portfolio diversification is or is not sufficient to make “downside” and “upside” aspects of portfolio volatility of comparable magnitude.
In this presentation, we will explain an intuitive way to understand this problem based on the contingent claims approach to credit risk (Merton 1974) and provide statistical tests for the symmetry assumption.
Finally, we will demonstrate how the Northfield Everything, Everywhere model and Optimizer automatically solve this issue by explicitly considering higher moment properties of bonds (i.e. negative skew and positive excess kurtosis) in portfolio risk estimation.
Use and Misuse of Effective Tax Rates
Presenter: Dan diBartolomeo
November 14, 2023
Abstract
When considering the effect of taxes in portfolio management, we must make a careful distinction between statutory tax rates and effective tax rates. For example, investments within a tax deferred account (e.g. a 401K in the US) would have a statutory tax rate of zero while the money is in within account, but would be taxed at ordinary income rates when withdrawn.
For long horizon decisions we need to estimate an effective tax rate over the life of the investment holding period between now and a future withdrawal. There are also situations where the implication of effective tax rates is that the difference between short-term and long-term capital gains is smaller than statutory rates would suggest.
Finally, we will examine how effective taxes can be incorporated into taxable asset allocation (see diBartolomeo, Horvitz, and Wilcox 2006; Markowitz and Blay 2016). It should be noted that improperly framed use of effective tax rates in rebalancing existing portfolios can lead to perverse solutions.
Capturing Climate Risk for Coastal Commercial Real Estate
Presenter: Emilian Belev
October 26, 2923
Abstract
The adverse impact of climate induced events manifest itself with an increasing frequency as damages caused to commercial real estate near the coastline.
In this presentation we make an overview of an actionable model with which investors can incorporate the probability and severity of such events in the overall risk analysis and asset selection process of their real estate portfolios.
The method leverages data readily available from public sources, the Northfield real estate risk model, and innovative statistical techniques to produce results of practical significance to investors, lenders, insurers, local governments, and climate regulatory bodies.
Performance Analysis of Market and Factor Timing in Active Equity Management
Presenter: Dan diBartolomeo
October 12, 2923
Abstract
It is routine for quantitative strategies to be analyzed based on the per period active returns associated with “factor bets” that are present in the portfolio strategy. At the broadest, one could think of the return of a benchmark index being associated with the “market” factor. Routine factor-based attributions are done on a period-by-period basis (day, month) and then aggregated over time to make statistical assertions such as “your low P/E tilt added X basis points per month of return over the sample period.”
However, to fully understand the implications of such results, we really need to get even further granularity in the analysis. We need to understand whether holding active factor exposures constant would have produced a better or worse result, as opposed to having factor exposures vary over time. It is our experience that many factor based strategies (i.e. a value tilt) produce materially reduced active returns because the factor exposures vary through time in detrimental ways. Put another way, the alpha contribution of factor timing is negative even when the alpha contribution for factor tilts is positive.
In this webinar, we will present a live case study using output from our performance attribution system. We will show that the opposing signs on the alpha contribution of factor tilts and factor timing is often to be expected when viewed in the “portfolio dynamics” framework proposed by Sneddon (Journal of Investing, 2006) whenever transaction costs are non-zero.
Climate Change, Real Estate and the Bottom Line
Presenter: Simon Camilo Büchler, Research Scientist and Director of the Price Dynamics Platform at the Massachusetts Institute of Technology (MIT) Center for Real Estate
September 28, 2023
Abstract
Commercial real estate (CRE) is a crucial asset class (Ghent, Torous, and Valkanov, 2019). According to MSCI Inc., in 2020, approximately US $10.5 trillion of global real estate assets were under institutional management for investment purposes. Nareit estimated that the total value of CRE in the U.S. in 2018 was $16 trillion. Additionally, pension, endowment, and foundation funds controlled over $9 trillion in total assets, with almost $800 billion invested in real estate as of 2019.
Despite the significance of CRE as an asset class, we know little about how climate risks impact its performance.
This webinar will review the recent academic literature on climate risk and real estate.
Dealing with Super-Concentrated Positions in Taxable Portfolios
Presenter: Dan diBartolomeo
September 14, 2023
Abstract
Private wealth managers are often faced with the unique challenge of a client whose current equity portfolio is often super-concentrated into just one very large holding. Such situations often arise from a private company being merged into a publicly traded company. The founders of the private firm often find themselves owning massive positions of a traded firm at a low-cost basis in which they no longer have a controlling interest. Liquidating such a position immediately would incur very large capital tax and so is considered an unattractive approach to achieving a reasonably diversified portfolio. A similar problem is faced by many family offices where there may be very large positions in three or four public companies where family members have been founders or senior managers.
In a recent webinar, we demonstrated methods for transition of a taxable legacy portfolio to a new composition by budgeting the tax costs within and across tax years.
In this presentation, we expand this theme to the “complementarity portfolios” which include the use of both margin borrowing, derivatives, and unusual optimization parameters to supercharge tax loss harvesting and greatly accelerate the transition process. Our real-life example is a taxable portfolio consisting of $600 million dollars in a single stock starting at zero cost basis.
Marking-to-Market Private Credit
Presenter: Emilian Belev
August 30, 2023
Abstract
Private Credit expansion as an asset class for institutional investors accelerated throughout the years leading up to the COVID pandemic. That growth was based on the search for diversification and yield by investors that looked for a risk profile between that of public markets and private equity. Another important factor for its popularity was the provision of capital for levered loans for buyout deals and other lending to private equity managers seeking to boost their IRR by delaying capital calls in the early years of their funds.
Post-pandemic Private Credit has resumed its upward trajectory in terms of popularity, but for different reasons. In the absence of attractive exits, private fund managers look for other ways to maintain the scale of their business, and without the headwind of expansive monetary policy and easy bank lending the end of which was denoted by increasing interest rates, providing private loans prove to be a more attractive investment.
Given its increasing importance, it is essential that investors gain an insight of the true value of positions in the asset class at any point in time. The presentation will focus on the methodology and application of the valuation of private credit investments to create historical time series of their mark-to-market (MTM) return and, consequently, the analysis of such time series in a risk factor model. The eventual result is the trifecta of the investment process by enabling three distinct functions - valuation, risk, and performance - independently and in addition to the GP-supplied information for the same purpose.
The results support a strong case for the relationship between MTM private credit returns and the Northfield multi-asset class risk model.
Expressing Alpha on Non-Linear Derivatives
Presenter: Dan diBartolomeo
August 15, 2023
Abstract
Traditional derivative pricing models such as the Black-Scholes model for options have a feature that many investors find unintuitive. The investor’s expectations about the expected return of the underlying asset (e.g. the “alpha” of a stock) are not part of the inputs to valuing options. Common intuition might be that if the “alpha” on a stock was positive, that would make call options more attractive than put options, which should be reflected in the respective option prices.
In this presentation we will briefly dissect the assumptions under which derivative pricing models operate such that underlying alpha is presumed irrelevant. We will also provide appropriate procedures for expressing the alpha on an option conditional on the alpha of the underlying security.
Estimating the Probability of Equity Market Crashes Portfolios
Presenter: Dan diBartolomeo
July 27, 2023
Abstract
Among both investors and finance academics, the VIX volatility index has become known as the equity market’s “fear gauge.” During rare periods of crisis, the VIX estimate of S&P 500 annualized volatility often exceeds 80% (four to five times the historical average level). As investor’s fear loss rather gain, using a symmetric measure is indirect at best.
In this presentation we will illustrate how to use the VIX to decompose equity market risk into two states to directly address the probability of large loss.
In the first state, the market goes into the future with typical levels of annual volatility (e.g. 15-20%).
In the second state, we will assume the equity market crashes with a crash defined as a negative return that is an N (e.g. 3) standard deviation event relative to typical volatility levels.
With these attributes defined, we will illustrate how to estimate the probability of a crash within the investor’s chosen time horizon.
The underlying mathematics are related to the “Generalized Currency Risk Model” provided in a recent webinar. Also used is methodology from diBartolomeo (Investments and Wealth Monitor, 2020) which dealt with forecasting how long high market volatility would persist at the outset of the COVID pandemic.
Optimal Wash Sale Behavior for Taxable Portfolios
Presenter: Dan diBartolomeo
July 13, 2023
Abstract
The management of taxable portfolios is made far more complex by the presence of “wash sale” rules in the tax codes of the United States and several other countries (e.g. United Kingdom). These rules disallow an investor from deducting a capital loss on their tax return for the year in which the loss occurred. Some taxable managers are adamant about avoiding any disallowances, while other managers will tolerate wash sales when there are compelling investment reasons for such a transaction.
While not intended to be legal advice, this presentation will discuss the general structure of the US wash sale rules, and possible exceptions to the common understanding of these rules by investors. The presentation will then discuss how wash sale issues have been incorporated into the Northfield tax optimization algorithms. To the extent that managers have different degrees of willingness to incur wash sales, the Northfield Optimizer has considerable flexibility to provide varying degrees of wash sale avoidance.
The latter portion will present a simple framework that a manager can use to determine the wash sale behavior that is mathematically optimal for their clients, conditional on a set of inputs. Those inputs include expectations about future tax rates, current unrealized gain/losses, market returns, and limitations on tax loss-carry forward.
Finally, we will show how optimization inputs can be adjusted to implement the now-defined optimal behavior.
Generalized Currency Risk and the Existence of a Risk-Free Asset
Presenter: Dan diBartolomeo
June 27, 2023
Abstract
The recent congressional gamesmanship over the US Federal debt ceiling leading to a possible sovereign default in the world’s reserve currency illustrates the high degree of geopolitical influence routinely impacting major currency markets.
At the other end of the spectrum, Northfield recently added “on demand” asset coverage of more than one-hundred “frontier market” countries, where political and economic stability are fragile (e.g. Sudan).
In this presentation, we will first illustrate our updated methodology for estimation of currency risk in a variety of situations arising for international investors across the range of "floating," "managed," and "pegged" central bank regimes. The central portion of the presentation will be devoted to examining the implications of a sovereign default (even of a technical nature) in the US dollar as the world’s functional reserve currency. Perversely, our modeling suggests that a sovereign debt default by the United States would strengthen the US dollar relative to other currencies in an ensuing crisis, as the US would remain the “safest haven” in a world left without a risk-free asset.
The final part of the presentation will be devoted to discussion of how traditional asset pricing theory (e.g. CAPM) would change in the absence of a recognized risk-free asset.
Private Credit Part 2: Commercial Real Estate Debt
Presenter: Emilian Belev
May 30, 2023
Abstract
In this presentation, we focus on estimating the probability and severity of default of CRE-backed debt using real estate risk models in combination with property or portfolio-level cash flow and fair value forecasting. As is known, Northfield was a pioneer in developing a risk model for commercial real estate close to two decades ago. Since then, we have increased our coverage to more than 2000 global markets. This expertise combined now with the cash flow and valuation forecasting technology developed by our partner Aspequity, gives us the ability to estimate the probability of default and expected loss from default due to both a shortfall of periodic cash flows from a real estate investment with respect to periodic debt service, as well as property value shortfall with respect to the loan balance. We analyze this application in the context of various mortgage structures like interest-only vs. fully amortized, fixed rate vs. floating rate, etc.
In addition to the importance of the direct application of this model in the current environment, which is particularly challenging for the real estate asset class, the framework is also well suited to perform stress tests and scenario analysis.
A Plausible Short Cut for Tax Optimization and Why NOT to Use It
Presenter: Dan diBartolomeo
May 16, 2023
Abstract
In the nearly thirty years since capital gain tax functionality was added to the Northfield Optimizer many firms have tried to imitate our process, while taking a variety of short cuts in order to simplify the computational and conceptual complexity of the problem. Most of these short cuts can be summarized as “optimize the portfolio on a pre-tax basis, while constraining tax dollars to be paid in the current tax year to a chosen maximum.”
In this presentation, we will demonstrate how to use this shortcut method in the Northfield Optimizer, but also illustrate why this simplification provides inferior outcomes for taxable portfolios, irrespective of whether they are active or passive investors.
Private Debt Risk - Part 1: Private Commercial Lending
Presenter: Emilian Belev
April 25, 2023
Abstract
While nominally representing the same type of transaction and stake in the payoffs of an enterprise, private and publicly traded debt have some notable differences in terms of their risk-return characteristics. The differences originate from the fundamental nature of these two types of investments. The ability to trade public debt continuously defines a different role in investor portfolios, with different time horizons and clientele, as well as defining risk factors, in contrast to non-traded private debt. By extension, the lock-up character of private debt has influenced the terms and conditions of such loans to favor floating rate interest and more flexible reset provisions.
In this presentation, we will look at private debt both from the perspective of direct lending and fund investing and investigate the appropriate risk metrics that reflect the true nature of such investments.
We will also point out some pitfalls which arise from attempts to force an off-the shelf public market risk framework on private debt due to insufficient understanding, flawed advisory incentives, or lack of an appropriate private investment analytics toolset by some organizations.
Navigating Investment Funds Manager Skill
Presenter: Dan diBartolomeo and Bill Pryor
April 18, 2023
Abstract
With tens of thousands of investment funds to navigate, it is often an overwhelming challenge to identify those funds which both align with a desired investment objective and whose managers consistently demonstrate skill in investment management. Using publicly available fund data, Northfield developed the theory and methodology to determine a fund's expected excess return arising from manager skill as a forward-looking performance indicator, the Precision Weighted Excess Return (PWER). This is compared to fund rating services that rely on less nuanced, backward-looking processes.
In a recent collaboration, Northfield tasked the investment data and analytics firm Investics to provide a publicly available online service to interrogate, screen and visualize this manager skill framework, named Investment Manager Evaluation Analytics (iMEA).
Northfield President Dan diBartolomeo will review the origins of the theory which had its first practical use in the Bernie Madoff investigation in 1998. Investics President Bill Pryor will review the process of modeling and enrichment of the manager skill data, followed by a short demonstration of the online tools.
We believe this new tool has wide application in manager selection, performance evaluation, and fund marketing.
Banking Crisis in Review: Implications for Public and Private Assets
Presenter: Dan diBartolomeo
March 30, 2023
Abstract
Global financial markets have been inordinately volatile in the past couple weeks. The collapse of Silicon Valley Bank into the control of regulators, and the very public rescue of global bank Credit Suisse by the Swiss National Bank have raised obvious investor concerns about the potential for another banking crisis on the scale of the so-called Global Financial Crisis.
In this webinar we will examine these troubled banks from the perspective of proprietary Northfield analyses. These Northfield analyses include our Risk Systems That Read® risk estimates, our internal metric of corporate sustainability and our proprietary credit ratings.
We will then extend the analysis to the US and global banking industries to evaluate the potential for the current stresses to blossom into a full-fledged global crisis.
Finally, we will discuss the “two way” interrelationships between the banking system and other asset classes including sovereign debt, developed market equities, emerging markets, private equities, corporate bonds, and real estate.
Advanced Optimization with Downside Risk
Presenter: Dan diBartolomeo
March 14, 2023
Abstract
Since the beginnings of portfolio optimization in the 1950s, there have been cases where investors choose not to look at risk as volatility of return around the expected mean, but rather as “downside” volatility in terms of underperforming a specified level of return. In many cases, this chosen level of return is the minimum acceptable to keep the investor financially sound (see Roy, 1952 and Sortino, 1994) or as the expected return being used by actuaries in making financial projections for pension funds and insurance companies.
In this presentation, we will illustrate two ways to structure “downside” risk cases in the Northfield Optimizer.
Asset Allocation for Investors with Liabilities
Presenter: Emilian Belev
February 28, 2023
Abstract
Modern portfolio theory has centered on optimization over a single horizon where investor preferences are often described only cryptically through a risk aversion parameter. This popular framework, however, falls short of the real needs of many large institutional investors that have an identifiable stream of liabilities over multiple annual horizons - pensions, insurance companies, endowments and foundations, etc.
In this presentation, we demonstrate a robust approach to a multi-period optimization of a multi-asset class portfolio where the investor is maximizing the long-term portfolio value subject to minimum level of liquid resources available in each interim period to meet required liabilities. The approach is particularly appropriate for portfolios containing private assets that produce cash flows but are not themselves available for sale due to their illiquid nature.
Joint Consideration of Market Risk and Operational Risks
Presenter: Dan diBartolomeo
February 14, 2023
Abstract
The recent implosion of the FTX cryptocurrency exchange has reminded investors that there are two kinds of risk for investors. There are always elements of both market risks and operational risks to every investment process. In most cases the economic magnitude of operational risks are greatly minimized through the use of organized securities exchanges whose transactional integrity is guaranteed through a clearing operation that is independently capitalized. In addition, the general solvency of financial institutions has been de facto supported by governments and central banks as in the “bailouts” of the Global Financial Crisis of 2007-2009. However, investor losses associated with the collapse of Lehman Brothers and other financial institutions during the GFC were certainly substantial. There are several common investment situations such as “over the counter” derivatives and digital assets where calculations of financial risk assessments where the joint consideration of market risk and counterparty risk is necessary, and sometimes required by regulation.
In this presentation we will provide a computational framework to combine market risk and operational risks into a single numeric magnitude to be used in risk assessment and decomposition. We will illustrate the methodology using the example of the digital asset token called a “stablecoin” that were designed to have fixed dollar values (i.e. no market risk) but where design flaws in their financial engineering amplified operational risks to the point of widespread collapse.
Beyond Black-Litterman: Capital Market Assumptions with Heterogeneous Investors
Presenter: Dan diBartolomeo
January 31, 2023
Abstract
The Black-Litterman model (Journal of Fixed Income, 1991) was an important milestone in providing investors with a practical approach for blending their own views of asset class returns with a set of independently derived prior beliefs. Internal “views” of capital markets are often presented as a model asset allocation rather than explicit numerical forecasts.
In the case of BL model, the assumption is that global financial markets are efficient so that expected returns for various asset classes can be inferred for a given asset class covariance matrix. If we assume that the covariances of each asset class are sufficiently defined by the beta of the asset class to the world wealth portfolio, we are employing the equilibrium inherent in the International Capital Asset Pricing Model (ICAPM) as first proposed in Solnik (1974). While ICAPM does capture global currency of denomination effects (i.e. home market bias), most of the simplifying assumptions of the original CAPM are left in place. Those assumptions omit many of practical aspects of the asset allocation problem across heterogeneous investors. For example, the preferences of a tax-exempt defined benefit pension are likely to be very different than the preferences of family office of taxable investors. Among these points of likely distinction are liabilities to fund future consumption, risk tolerance, regulatory constraints, leverage constraints, taxation and the preference for income versus capital growth.
In this presentation, we will present a model which parallels Black-Litterman in most ways, but explicitly provides for heterogeneous preferences in portfolio formation.
Realistic Scenario Analysis and Stress Testing with FASST
Presenter: Dan diBartolomeo
January 17, 2023
Abstract
Many investment organizations try to supplement their regular risk models with “scenario analysis” and “stress testing” as originally developed for asset/liability management of banks and insurance companies. Unfortunately, the traditional methods embed many unrealistic assumptions (e.g. macro events like the Global Financial Crisis occur instantaneously) such that the analytical output is of minimal value, as pointed out in two Northfield newsletter articles in 2006. Unsurprisingly, getting useful output from such efforts requires that the input scenarios be realistic.
In this workshop, we will illustrate how to formulate scenarios and stress tests that are sufficiently realistic to be useful. The first requirement is that stress scenarios be of reasonable probability. A second is that uncertainty in the scenario element be addressed (e.g. “Oil prices will rise between 10 and 30% over the next six months, as opposed to oil prices will rise exactly 20%). The third is accounting for serial correlation in conditioning variables (e.g. inflation and interest rates trend rather than being random walks).
With a set of realistic inputs in hand, Northfield clients can use our FASST system to produce very rich analytical output. The FASST system is an extension of the method proposed in Meucci (“Fully Flexible View”, 2008). Unlike traditional scenario methods where the outcome of each element of the scenario is deterministic, FASST output is a forecast of the entire multi-period return distribution conditional on each element of the scenario (e.g. “How will my portfolio perform if oil prices go up between 10-30% in the next six months?”). The multi-period FASST analysis can cover all time horizons from intra-day for a trading desk to a decades long horizon appropriate to asset/liability analysis for a pension fund.
Valuation of Private Companies Using Risk, Growth, and Time
Presenter: Emilian Belev
December 22, 2022
Abstract
Private companies are long-term investments that generate return over multiple years in the form of cash inflows. The main characteristic of these investments is that they cannot be readily sold in the open market at any particular time in the future. Therefore, their economic value is derived almost exclusively from the ability to generate cash flows over the firm's life.
By using custom growth-rate assumptions, a Northfield equity risk model, and a model for the evolution of cash dividends, we utilize the EXPLO valuation model to first generate the statistical distribution of the cumulative future value of dividends over the economic life of the company, and then to estimate the current fair value of the company using market derived level of loss aversion.
Special attention will be paid on the business stage of the company as determinant to all three-input model parameters, and the factors that impact the growth of a company.
Tax Efficient Management for Traditional Mutual Funds
Presenter: Dan diBartolomeo
December 13, 2022
Abstract
Since the introduction of tax sensitive portfolio optimization at the lot level by Northfield in 1995, there have been significant advances to maximize after-tax risk adjusted returns in separately managed accounts. Over the same interval, both passive investing in general and ETFs have greatly eroded the market share of traditional mutual funds. A recent class action lawsuit against Vanguard for poor tax management of one of their funds has focused even greater interest on the issue.
This presentation will illustrate what techniques can be usefully employed by open end mutual funds to be optimally “tax smart,” as obviously preferred by taxable investors.
The US SEC already requires mutual funds to publish after-tax returns as part of marketing material, a fact of which investors are increasingly cognizant. Among the challenges are that relative to traditional mutual funds, ETFs have legal methods to make themselves more tax efficient, in addition to the low turnover associated with generally passive management. In the US, funds are subject to more restrictive treatment of capital gains and losses. Almost all mutual funds have heterogeneous shareholders that include both taxable and tax-deferred investors (e.g. 401K assets).
Finally, mutual funds frequently have daily cash flows from investors adding money to the fund or making withdrawals. In turn, the fund is required to make daily trades which often necessitate violations of “wash sale” rules.
Scenario Analysis and Stress Testing for Private LP Funds
Presenter: Emilian Belev
November 29, 2022
Abstract
The lack of analytical tools that link private fund cash flows and NAVs with real world risk factors and macro variables has long put private fund investment teams at a disadvantage with respect to their public asset class peers when it comes to scenario analysis and stress testing. Northfield has offered pioneering analytics for both scenario analysis and forecasting private fund cash flows and NAVs.
In this session, we will demonstrate how these tools effectively combine to allow private fund investors to analyze the full range of scenarios of their choice.
The first step in this process is to bootstrap a statistical distribution of portfolio growth vectors that have been historically exhibited under the conditions defined by the user specified scenario.
The second step is to provide the growth vector distribution as an input to the cash flow and NAV simulation engine to generate highest likelihood forecasts for the private fund performance under the set scenario environment, as well as likelihood ranges at different levels of confidence.
This process can also be extended to reverse stress testing which involves finding the worst scenarios that can hurt the portfolio with a material likelihood level - a probability of occurrence being an input from the investor - and then evaluating a forecast under these conditions. In this way, hypothetical events that are severely adverse but have very low likelihood of occurrence in terms of joint probability of variable outcomes do not misrepresent that actual material risks faced by the investor.
A Rules-Based System to Get the S&P 500 to Outperform Itself
Presenter: Jason MacQueen
November 15, 2022
Abstract
Suppose your investment goal was to earn the Market Risk Premium, but, like many investors, you would also like some outperformance without taking extra risk. Would you be happy with a fund that generated average annual outperformance of +0.72%, over a 15-year period from 2007 to 2021, with the same or lower risk as the S&P 500, and with turnover of less than 7% p.a.?
According to Modern Portfolio Theory and the Capital Asset pricing Model (CAPM), the market portfolio is supposed to be efficient (in the Markowitz sense), and all stocks should have the same risk-adjusted expected returns.
Crucially, it is this assumption of market portfolio efficiency that justifies active managers using capitalization-weighted indices as benchmarks in optimization. However, if the benchmark is inefficient, then optimizing against is likely to bake the same inefficiencies into the portfolio, thereby making it harder for the manager’s stock selection skill to affect the portfolio’s performance.
The S&P 500 is widely regarded as a reasonable proxy for the market portfolio, so we can think of the SPDR as a capitalization-weighted factor ETF intended to capture the market risk premium. As we have already shown1, however, capitalization-weighted factor ETFs do not always do a very good job of capturing a targeted factor premium.
Starting with CAPM expected returns, and using a Rules-Based System, we identify which holdings in the capitalization-weighted S&P 500 are most likely to be inefficient, and adjust them to make them more efficient. By eliminating some of the unintended bets that obscure the market risk premium performance, we get a more efficient portfolio (N.B. same stocks, different weights) with higher return and the same or lower risk, hence showing that the S&P 500 cannot be an efficient portfolio.
In this talk, we will give several examples of such a strategy.
The $17 Billion Question: Two Spectacular Failures in Risk Management
Presenter: Dan diBartolomeo
October 27, 2022
Abstract
In last year’s Archegos fund debacle, five major investment banks lost an aggregate of over $11 Billion on failed margin calls. This was a spectacular failure in risk management at what are purportedly the most sophisticated financial institutions in the world. A similarly horrific outcome arose a few years ago at JP Morgan during the so-called “London Whale” incident (loss $6.5 Billion).
While these events are popularly ascribed to “operational risks from procedural failures,” a deeper analysis of published investigative documents suggests that common flaws in the analysis of both market risk and liquidity contributed greatly to the severity of the negative outcomes despite the availability of appropriate analytical methods.
The final part of the presentation will be devoted to describing a clear analytical “shortcut” that has become pervasive across the asset management industry. This error has the potential to be the next cause of similarly dire results for certain typical fund portfolios..
Advanced Tax Optimization
Presenter: Dan diBartolomeo
October 13, 2022
Abstract
Northfield's Optimizer provides many sophisticated tax management features. Many users are not familiar with these capabilities and hence produce suboptimal outcomes.
In this presentation we will demonstrate how to use these advanced features to address situations like tax gain harvesting, negative tax rates, qualified versus non qualified dividends, short positions, and inclusion of derivatives.
A final focus will be proper risk tolerance setting for active management and direct indexing.
Measuring Exposure for LP Funds
Presenter: Emilian Belev and Thomas Meyer
September 29, 2022
Abstract
Measuring the exposure to limited partnership funds investing in private assets is a key challenge to multi-asset class investors. It arises from the long time-lags between milestone events: the commitment to the fund, the capital calls, and returning capital to the fund’s investors. The task is further complicated by the dual perspective of the portfolio of the fund as seen through the prism of the end investor or the manager.
This work explores the variety of existing and potential measures that address this challenge, comparing their appropriate usage and potential adverse effects. The goal is to provide a multi-faceted view to help improve investment decision-making.
Quantifying Economic Narratives with Media Attention - A New Approach to Asset Pricing
Presenter: Ronnie Sadka
September 13, 2022
Abstract
This webinar introduces a media-attention approach to quantifying economic narratives in a systematic fashion. Asset-price risk exposures are measured for hundreds of attention-derived narratives, including macroeconomic, ESG, crypto, and emerging trends. Narrative betas can be used to gain or hedge portfolio exposure to narratives by tilting portfolio weights through optimization or by constructing dedicated basket portfolios of narrative-sensitive assets. Emerging narratives are systematically identified---portfolios of trending/emerging narratives tend to outperform in the short-run.
The framework introduced here extends the traditional paradigm of factor models to a more general class of intangible factors, combined with behavioral elements pertaining to investor attention.
Active Returns from Passive Management: Cointegration of Sector Returns
Presenter: Dan diBartolomeo
August 30, 2022
Abstract
To a large extent, changes in the profitability of publicly traded companies are economic transfers from one sector to another. For example, increases in energy prices make energy producing companies more profitable but make energy consuming companies less profitable. Taken together, the total profits of the two sectors may be stable while the profits of each sector viewed separately would be volatile.
To the extent that a combination of multiple time series has the statistical property of stationarity when the underlying series are not stationary is called cointegration.
In this presentation, we will illustrate a simple way to use cointegration methods to better forecast long term factor returns and define sector weights that have the maximum likelihood to outperform conventional equity indices over long horizons.
Betting on Volatility: Variance Swaps and the Hall of Warped Mirrors
Presenter: Emilian Belev
August 11, 2022
Abstract
In this presentation which lies at the cross-section of modeling theory and practice, we examine Northfield’s capability in the higher-end of the derivative analytics spectrum – Variance Swaps. At a time where divergent calls of the steepness of an upcoming recession are a fertile ground for betting on volatility, we accentuate that the instruments used to make such bets represent a non-trivial task in a broad multi-asset class risk model. The challenges come from strictly non-normal outcome distributions of variance swaps.
We will examine two key relationships which make modeling risk of these investments tractable: the connection between continually forecasted volatility and implied volatility, and the link between implied volatility and the outcomes of the underlying asset. A model that results from combining these two insights not only allows for the estimation of linear exposures to risk model factors, but also naturally incorporates the calculation of the skew and kurtosis represented by these investments.
A summary observation is that variance swaps represent a unique combination of two compounding layers of leverage, and therefore should not be undertaken unless the risk impacts are well understood.
Valuing Liquidity: Estimating the Price of the Option to do “Something Else”
Presenter: Dan diBartolomeo
July 28, 2022
Abstract
In recent decades, a broad range of institutional investors have dramatically increased their allocation of capital to illiquid assets such as private equity, real estate, and financing of public infrastructure. This trend is typically justified by asserting an expectation that some investors have unpredictable consumption patterns and therefore a willingness to accept lower returns for high liquidity. It is assumed that investors who have highly predictable consumption expenditures (e.g. a pension fund) can therefore obtain higher returns with little disutility by giving up liquidity that they believe they will never need. However, having a high proportion of illiquid assets means that the investor has given up the ability to change their asset allocation should they wish to do so, or even to rebalance their allocation back to previously defined weights as the market value of portfolio assets fluctuate through time. As compared to another investor with similar but liquid assets, the illiquid investor is now short the “option to do something else.”
In this presentation, we will present an analytical model for this option which provides a more appropriate estimate of the “illiquidity premium” that is specific to the investor’s overall portfolio and the certainty of the time horizons for their consumption liabilities. Like most option pricing problems, the key input to the model is the volatility of the underlying which in this case is the cross-sectional dispersion of returns across asset types.
Estimation Error and What to Do About It in Northfield Tools
Presenter: Dan diBartolomeo
July 14, 2022
Abstract
This online workshop will focus on using Northfield methods for addressing estimation error bias in portfolio construction. When we form portfolios, we naturally want to optimize return/risk tradeoffs as first proposed in the early Markowitz papers (1952, 1959). However, such optimization has been criticized as impractical for decades as the required inputs are the known probability distributions of the asset returns, and the correlations among those distributions. Since the asset returns of interest are in the future, we cannot know the return distributions. We only have estimates of what the returns distributions will be. To the extent our estimates of future return distributions can be wrong, we need compensate for that possible wrongness in our expectations of portfolio return and risk.
Many different methods have been proposed as ways to resolve estimation error in portfolio construction. We will illustrate the three algorithms that Northfield software users can access to adjust portfolio construction for estimation error. The two techniques focusing on return are the method of Jorion (1985) and that of Black and Litterman (1990). The third algorithm, Ledoit and Wolf (2004) focuses on possible estimation error in return covariance.
We will also discuss how many aspects of estimation error are effectively built into Northfield risk models to minimize the extent of possible bias.
Are Equity Indices Actually Diversified?
Presenter: Dan diBartolomeo
June 30, 2022
Abstract
One of the most basic assumptions of modern investment practice is that equity indices often used as benchmarks (or passive portfolios) are inherently well diversified. This assumption arising from a number of early research papers asking “How many stocks does it take to make a diversified portfolio?” The answers varied widely from ten in Evans and Archer (Journal of Finance, 1968) up to around forty Statman (Journal of Finance and Quantitative Analysis, 1987). However, these papers all use different definitions of what constitutes a “well diversified” portfolio in a single asset class context (e.g. is my portfolio sufficiently like the market?).
This is presentation will introduce a proprietary measure of diversification that captures the correlation among securities in a portfolio measured in absolute returns, which is more appropriate in a multi-asset class portfolio. Our new measure is expressed in the number N (e.g. 20) of equally weighted positions that would be comparably diversified, assuming each security had the average level of volatility of all members of the index, and all securities are uncorrelated.
Empirical examples will illustrate that due to capitalization weighting and correlation across securities, many popular indices that contain hundreds or even thousands of securities are far less diversified than investors presume.
Using Northfield Tools to Distinguish Between Tracking Error and Active Risk
Presenter: Dan diBartolomeo
June 14, 2022
Abstract
The asset management industry generally uses the terms “tracking error” and “active risk” as being equivalent for both passive and active strategies. This is untrue for an obvious reason. Every active manager (and investor using an active manager) must believe ex-ante that their strategy will produce positive benchmark relative returns. If the benchmark represents the opportunity set for the average investor, it is arithmetically impossible for everyone to produce an above average return, just as there can be no class in which all students have above average grades. A large fraction of active managers must therefore be wrong about the expectation of positive outperformance. As a metric of risk, tracking error does not account this unavoidable “wrongness.”
This workshop will explain the alternative calculation of “active risk” which is provided in Northfield risk reports. We will also demonstrate an even more sophisticated way to describe the distinction, which implies that the industry practice of trying to maximize “information ratios” for active portfolios is ill advised even in theory.
Rationalizing Equity/Bond Market Correlations in Asset Allocation
Presenter: Dan diBartolomeo
May 26, 2022
Abstract
One of the most basic questions in asset allocation is the expected correlation of equity and bond markets. Estimating a reliable expected value of this key input is often elusive. Even empirical studies based on purely historical data often disagree on the sign as well as the magnitude.
In this presentation, we will examine several of the structural issues that can be used improve the asset allocation processes.
First, we will consider the legitimacy of the assumptions which underly the use of a correlation coefficient in the first instance. Correlation is conceptually presumed to describe the relationship between two random variables, but there is conclusive empirical evidence that bond market returns are not random. Interest rates (and bond returns) can exhibit both negative and positive serial correlation due to the influence of central banks. This results in biased estimates as the traditional Pearson correlation measure is not robust. Whether you observe positive or negative correlation between equities and bonds is strongly impacted by the frequency observations (daily, monthly, annual). In addition, a precise definition of what the bond market consists of is necessary both in terms of the issues included and differences in composition over time. If the “bond market” includes corporate bonds you will naturally get more positive correlation than for sovereign bonds, as Merton (Journal of Finance, 1974) shows that a corporate bond can be decomposed into a riskless bond and equity of the issuer. There must obviously be a correlation approaching one between observable equity returns and equity returns implicit in corporate bonds.
Finally, we must separately consider the cases of sovereign debt in a global reserve currency (e.g. US$) and bonds of a typical government. The long-term creditworthiness of sovereign debt is tied to underlying economic activity of which the local equity market is a forward-looking indication, as described in Belev and diBartolomeo (Contingent Claims Analysis in Corporate Finance, Crouhy and Galai editors, 2019).
Optimal Turnover: The Tradeoff Between Alpha Decay and Transaction Related Costs
Presenter: Dan diBartolomeo
May 12, 2022
Abstract
In this webinar we will describe methods for finding the optimal level of portfolio turnover for an active strategy. This involves tradeoffs between optimization inputs such expectations of alpha values and alpha decay, transaction costs, and capital gain taxes (if applicable). The usual mean-variance objective function is in units of risk-adjusted expected return per unit time, while transaction costs and the liability for taxes arise at specific moments in time.
The Northfield Open Optimizer allows users to input an amortization scalar (implicitly defining a time horizon) so that costs are transformed into the correct units. At a basic level, the amortization scalar is directly related to the expectation of portfolio turnover. A more nuanced understanding of the optimal tradeoffs includes adjusting the amortization constant for the rate of alpha decay, the liquidity of the securities, and the active share of the strategy.
Finally, we will provide a method to explicitly approximate multi-period optimization within the single period structure of MPT.
Sustainability as a Portfolio Level Risk Measure
Presenter: Dan diBartolomeo
April 26, 2022
Abstract
The concept of corporate “sustainability” has become a key buzzword for investors in recent years. While there has been global effort to develop metrics of corporate responsibility (i.e. ESG measures) at the firm level, there has been relatively sparse research on how ESG concepts contribute to investment performance at the portfolio level.
In previous webinars, we have reviewed the literature on ESG/SRI effects on actual investment performance as inconclusive in terms of equity returns, but with relatively consistent data that firms with high ESG scores are perceived as less risky by lenders and hence pay lower interest rates on bonds and loans. In diBartolomeo (Journal of Investing, 2010) we introduced a quantitative measure of corporate sustainability which makes an inference on how long investors believe a particular firm will survive.
In this webinar, we will show the performance of quintile portfolios sorted by “the expected life of firms.” The empirical tests cover all stocks traded in the USA (i.e. including ADRs) for the thirty-year period from 1992 through 2021. We also examine two subperiods: first from the start of the data set to early 2010 and from spring of 2010 to the end of 2021.
The results suggest that quintile returns were not statistically significantly different across the “expected life” dimension early in the sample but grew highly statistically significantly different subsequent to the Global Financial Crisis (GFC, 2008-2009) and the start of the United Nations Principles on Responsible Investment (UNPRI, 2006). Volatility levels were monotonic and highly differentiated for all periods with the expected sign (lower risk in sustainable firms). Materially superior Sharpe ratios are observed for high sustainability quintiles (both equal and capitalization weighted), which is consistent with significant positive returns to “Quality” as reported in diBartolomeo and Kantos (Journal of Asset Management, 2020). To confirm robustness, all analyses were repeated with the entire financial sector removed to avoid potential bias arising from government “bailouts” during the GFC period. The sustainability measure is an effective metric to forecast the absolute volatility of returns of well diversified portfolios which can provide investors with materially better portfolio level Sharpe ratios.
Secure Data Aggregation on the Northfield NEXUS Online Platform
Presenter: Guest speakers Lloyd Oestreicher and Mike Oestreicher of Digital Financial
April 14, 2022
Abstract
A key element that is included in the new Northfield NEXUS analytical platform is a partnership with Digital Financial to provide industrial strength data aggregation with the highest available level of data security.
The Digital Financial network provides smooth connectivity between more than one hundred data providers including custodians, fund administrators, and vendors. For Northfield users, that data is delivered to a secure data warehouse (compliant with the SOC-2 industry standard) that has been integrated with Northfield online applications.
This workshop will illustrate how the DiFi OFX system not only streamlines the use of Northfield analytics, but also provides independent functionality including performance measurement and flexible hierarchical aggregation of portfolios for risk. Asset owners will also benefit from the ability to store user data on private assets (real estate, private equity, OTC derivatives) in data structures which facilitate the fullest use of the Northfield’s analytics for non-traded assets.
Finally, the DiFi capabilities include a “virtual asset master” application that can act as a critical interface between traditional back office systems and the use of optimization and risk applications on live portfolio holdings.
True NAV of Private Equity Funds - Alpha Generation by Limited Partners
Presenter: Emilian Belev
March 29, 2022
Abstract
Private equity fund NAVs are reported as an addition of NAVs of individual illiquid portfolio companies. In contrast to private funds, public fund NAVs values are derived from securities that can change hands from one fund to another overnight, which guarantee that their market value reflects the top marginal diversification benefits that such a security can provide.
Shares of private companies in a PE Limited Partnership fund do not have this feature as they do not get traded. At best, one can assume that GPs and their valuators use DCF models with discount rates based on market betas from public analogues. If there is undiversifiable risk in the PE fund portfolio, however, this approach makes the fund appear to have the same NAV as another fund with the same average company beta but one that is fully diversified, all else the same. Even if the first fund is riskier.
Conversely, market multiples (e.g. EBITDA to EV, or Sales to EV) could be used to calculate individual company NAV. In that case, the multiples are based on similar transactions. While some of the "comps" transactions may be acquisitions by publicly traded firms, the vast majority are buyouts. Therefore, the intrinsic value of such a transaction to one much larger fund in terms of diversification may not be the same as to another smaller one. Therefore, the market multiples may be misleading.
Over the long run funds with seemingly similarly reported NAV will have different cumulative realized value because of their different diversification level. This is based on the well-known fact that geometric return is adversely affected by higher interim volatility. Then, the ability to calculate a NAV value which properly reflects diversification becomes a source of alpha as it will distinguish between two funds of different diversification levels, all else the same. In other words, LPs can create alpha by combining existing investments with new investments, effectively increasing the NAV the existing portfolio by adding appropriately chosen new commitments that diversify the existing pool the most. That increase will be measurable and purely attributable to the LP's private equity team's investment decision.
Asset Manager Compliance with the SEC Rule 18f-4
Presenter: Dan diBartolomeo
March 10, 2022
Abstract
In just a few months, many US mutual funds will be required to provide a significant amount of new regulatory risk reporting under Securities and Exchange Commission Rule 18F-4. We believe that many asset management organizations are very ill-prepared to fulfill the requirements by the implementation deadlines. Fund managers must provide new forms of risk reporting and there are additional requirements for monitoring the quality of the risk estimation process by comparing realized drawdowns to forecasts of Value-at-Risk (VaR).
In this presentation, we will illustrate ways in multiple configurations for using Northfield’s risk models, data integration, and analytical applications address five key concerns in ways that ensure that both our clients and regulators have best possible process for meeting the new requirements.
To successfully comply with the new rule, asset managers need to overcome five challenges:
Inflation, Interest Rates, and Equity Valuation
Presenter: Dan diBartolomeo
February 24, 2022
Abstract
Equity markets have recently seen significant volatility associated with spikes in inflation across developed countries, and the associated expectations of interest rate increases by central banks.
In this presentation we will explore the dependence of equity prices to inflation and interest rates in two ways. First, we will consider the impact of various combinations of interest rates and inflation on the financial statements of two hypothetical companies. One will be a traditional company where the production function is immediate (e.g. a retail ice cream shop), while the other involves a long production cycle (e.g. manufacturing a cheese that must be aged). These illustrations will demonstrate that while real interest rates are a dominant ingredient to valuation, corporate tax rates and availability of borrowing also have influence. We will then extend these results into a Gordon style dividend discount model for theoretical confirmation that real interest rates should be a dominant factor.
The presentation will conclude with an analysis suggesting that equity valuations may be overreacting to near term changes in real interest rates while underreacting to changes in long term expectations of real rates.
The Alpha Lifecycle: New Research Into the Nature of Investment Alpha and What Portfolio Managers Can Do To Sustain It
Guest Presenter Chris Woodcock, Essentia Analytics
February 8, 2022
Abstract
In this presentation, we set out to validate and better understand an effect of return generation at the position level that has long been assumed but never demonstrated: that it has a life cycle -- a beginning, middle and end -- and that investors often hold on to positions too long, potentially diminishing whatever excess returns they were able to generate early on.
Our analysis examined roughly 10,000 “episodes” -- full cycles of a given position from first entry to last exit -- across 43 active equity portfolios over 14 years
The Other Road to Tax Efficient Investing
Presenter: Dan diBartomomeo
January 27, 2022
Abstract
In this presentation we will illustrate a forward-looking asset allocation strategy that can significantly reduce tax costs over the investor life cycle inclusive of major consumption events such as college tuition, retirement funding and estate taxes. Since Northfield created the first “tax aware” portfolio optimization system in 1995, most of the wealth management industry has slowly migrated from being “tax blind” to some degree of tax awareness, although often executed as relatively simplistic “tax loss harvesting” to reduce capital gain tax on equity portfolios.
However, an equally important source of “tax alpha” has been largely ignored which arises from the ability to avoid capital gain realization by reducing the need to revise or rebalance an investor’s asset allocation. For example, let’s assume an investor knows that they will need funds from a taxable account to pay for a planned consumption expenditure (e.g. college tuition) ten years in the future. One approach to funding the expenditure would be to wait until the date of the event and sell assets from the portfolio to obtain the required cash to cover the expenditure. That sale would usually require realization of capital gains resulting in tax costs.
An alternative approach would be to allocate investment income and investor cash flows as to build up enough cash in the portfolio to cover the planned expenditure at the required time. As the cash portion of the portfolio increases and decreases through time, the aggressiveness metrics (e.g. equity beta, bond duration) of the non-cash portion of the portfolio can be adjusted maintain the investor’s optimal allocation. The same process of managing cash flows minimizes the need to periodically rebalance the portfolio, thereby avoiding capital gain realization (i.e. you are usually selling what went up the most). Implementation of this strategy requires estimates of what the investor’s optimal allocation will be in future years. This concept is much like the “glide path” of a target date fund but customized to the individual household’s planned expenditures.
This process has been implemented in our WealthBalancer software with such forward allocations being obtained from the Generalized Logarithmic Utility Model of Rubinstein (Journal of Finance, 1976) as expanded by Wilcox (Journal of Portfolio Management, 2000 and 2003).
Reconciliation of Conflicting Risk Reports
Presenter: Dan diBartolomeo
January 13, 2022
Abstract
Investors use quantitative risk models to assess the magnitude and sources of risk within their portfolios. Unfortunately, risk like beauty is often “in the eye of the beholder” in the sense that different designs of risk models and factors can differ in their effectiveness for particular strategies (risk time horizon, absolute or benchmark-relative risk emphasis, diverse or concentrated portfolio).
For example, a Northfield client could use as many as eight different combinations of model design and time horizon for analysis of a US equity portfolio. Even greater diversity can exist across vendors. The specialization of models often leads to situations where the understanding of risk from different models may differ materially or actually be conflicting.
In this presentation we will present a practical approach to comparing the output of different risk models and reporting systems to obtain a robust understanding of the risk magnitudes, the uncertainty arising from estimation error and the usefulness of risk decomposition reporting.
Equity Style Factor Returns Revisited
Presenter: Dan diBartolomeo
December 30, 2021
Abstract
Empirical research into factor returns has long dominated the efforts of both quantitative equity managers and many academics. Whether done in the orthogonalized manner of commercial risk models (e.g. Northfield US Fundamental) or overlapping as typified by Fama and French (1992), factors are the lingua franca of modern equity investing.
Unfortunately, many investors fail to appreciate how factor return histories differ when viewed from the simple or orthogonal perspective, or when the factor returns are estimated conditionally on an asset pricing model (e.g. CAPM). These differences in perspective can entirely change an investor’s perspective of the returns associated with strategies (e.g. momentum, value, or low volatility).
In this presentation we will reconcile factor returns among several style factors as defined three different approaches: (1) the widely distributed Fama-French data, (2) the history of the Northfield US fundamental model and finally (3) a new version of the Northfield model that has been modified to reflect investor attention to large events (e.g. pandemics) as described in diBartolomeo and Kantos (Journal of Asset Management, 2020). Not only do some factors appear more important than others from different perspectives there are also important differences in return predictability and correlation across-factors.
Linking Value at Risk to Market Factors
Presenter: Mike Knezevich
December 16, 2021
Abstract
We traditionally consider Value at Risk or VaR to be part of the regulatory and policy reporting requirements as well as marketing for back and middle offices. While the front office normally employs a multiple factor risk model to estimate risk forecast and construct portfolios. As such, organizations incorporate two distinct separate measures with little connection, producing a disconnect between the various functions. A disconnect which hinders communication and feedback between the different “offices.”
For example, how can a portfolio manager make decisions on their portfolio while complying with the organization or portfolio’s stated risk objectives? Alternatively, how can risk measurement communicate to the managers where unnecessary risk is taken that impacts the organization’s stated risk level?
Northfield provides a solution by creating a link between VaR and Northfield’s multiple factor risk models. This solution allows users to better understand and communicate how the portfolio’s exposures to risk factors effects the portfolio’s Value at Risk. A connection that does not exist when using two different models for each measure.
As a risk measure, historical VaR is easily calculated using historical return data. This is adequate for risk measurement. However, an organization’s objective more frequently is to provide a risk forecast for investors and regulators who want to understand what risk will be in the future rather than what it has been in the past. In this case, historical VaR is a poor choice.
Northfield provides factor risk models which are specifically constructed to forecast risk. Although there are some differences between the assumptions of a multiple factor model and VaR calculations, with some adjustments the factor models produce a much richer VaR forecast. Calculating VaR based on a factor model adds new depth of understanding to the VaR forecast. There is now a direct connection to economically intuitive factors that explain market risk on which decisions can be made.
Some of these adjustments are made at the portfolio level while others are done at the individual security level. By applying the adjustments at the security level, we can use additional methodology and functionality that is not normally available using VaR to forecast risk. Methodology such as Northfield’s Risk Systems that Read (RSTR) which incorporates news and sentiment can now be embedded into the VaR numbers. Functionality such as optimization. By adjusting the risk with non-normality distribution parameters, we can now run an optimization while accounting for the non-normal of returns found in certain asset types.
EXPLO: A Unified Model of Investor Utility, Valuation and Liquidity
Presenter: Emilian Belev
November 30, 2021
Abstract
This presentation demonstrates a model of investor utility that builds on key principles of investor behavior that derives an intuitive mathematical result with direct real-world applications. It recognizes ideas familiar from prior work by Kelly, Rubenstein, and Wilcox, but reflects them in a distinct multi-period setting, incorporating the effect of leverage, periodic levels of required liquidity for consumption, and probability of bankruptcy.
In its basic form, the utility model shows that investors like expected profits, and dislike expected losses that are scaled by a loss aversion coefficient. The loss aversion itself is a function of leverage, and the amount and timing of required periodic liquidity. One of the implications is also that levered investors are particularly averse to kurtosis and negative skew due to the potential of bankruptcy. These ideas are intuitive to common sense, and the model provides the relationship among the variables involved.
An empirical study is included that reflects the application of the utility model as a valuation tool on three major public investments. The comparison of computed vs actual prices of these investments over two decades of monthly history demonstrates that the accuracy of the model to compute fair value is superior to other popular models.
Understanding Marginal Utility in the Northfield Open Optimizer
Presenter: Lalitha Raman
November 16, 2021
Abstract
The Northfield optimizer uses a gradient approach method to trade between two assets at a time moving a portfolio from its initial state to an optimal position.
This pairwise approach allows users to review the plausibility of each trade made step by step. Marginal Utility (MU) therefore is the catalyst for moving the assets in the direction of optimality.
This workshop is about gaining a better understanding of Marginal Utility and gives insight into how the optimizer constructs a portfolio.
Active Manager Skill
Presenter: Dan diBartolomeoh
October 26, 2021
Abstract
In recent years, there has been a large switch from active management to passive index funds by both institutional and retail investors. One plausible justification of this trend is that even if some active managers can persistently add value to investment outcomes, investors are unable to distinguish between active managers with such skills and those without.
This presentation will provide three different methods of identifying active manager skill and therefore the likelihood of persistent superior risk-adjusted performance in the future. All three methods will be available from Northfield and some of our partners. All three can be implemented by institutional investors or consultants as they do not require “inside the manager only” information.
Our first method, called “Precision Weighted Excess Returns” (see diBartolomeo and Warrick, 2005) involves only historical return data for the fund being evaluated and for a large set of similar funds needed for purposes of comparison. The PWER process involves statistical enhancements to increase the precision and accuracy of traditional evaluation of investment performance.
The second method is “Portfolio Opportunity Distributions” (PODs) as first proposed by Surz (JPM, 1994). In PODs, we need the historical record of a manager’s performance and a description of the limitations that were imposed on the active portfolio (security universe, benchmark, position size limits, liquidity constraints). Using a form of Monte Carlos simulation we can create a broad range of alternative portfolio returns that could have arisen had the manager made different choices but under the same constraints. By comparing the realized performance to the distribution of “what if the manager had done things differently” returns, we can quicky make statistically significant evaluations.
Finally, we will describe the method known as the “Effective Information Coefficient” (diBartolomeo, JPerfMeas, 2008) wherein we use risk models to estimate the implied security return values an active manager must have believed at each moment in time. By correlating the implied returns with the subsequent realized returns we can evaluate skill inclusive of all constraints. Given that the EIC analysis occurs at the security level (i.e. very large sample size) it achieves statistically significant results quickly but does require full transparency of past portfolio holdings.
Global Equity Model Alternative Perspectives
Presenter: Mike Knezevich
October 14, 2021
Abstract
Unintended risk can have an impact on portfolio returns. Managers construct portfolios with an in-depth understanding of the markets and assets in which they are invested. Many of their bets are well know and the risk is well managed. It is the unknown risk that goes unmanaged that can have a surprisingly detrimental impact on the portfolio’s performance.
When analyzing a portfolio, it is good practice to have a secondary view of the portfolio’s sources of risk to identify any unintended bets. Northfield provides various risk models allowing users to not only ensure consistent risk numbers across models but allows clients to gain a deeper perspective of their portfolio.
An extensive unbiased research project has proven that the hybrid approach to model construction is superior to other approaches to model building. Northfield’s constructs two different families of global equity risk models based on this methodology. The Northfield hybrid model is a combination of time series and statical factor modeling to create a best in breed risk model. Northfield’s double hybrid model combines cross-sectional, times series and statistical factors. Each is constructed differently to provide a unique analysis of portfolio risk. Together they provide a rich in-depth analysis of a portfolio using various factors.
In this product workshop we will discuss an analysis of a global equity portfolio using these two different Northfield global equity models. The high-level view using the Northfield global hybrid model and bottom-up view using the Northfield double hybrid models.
Building a Resilient US Equity Portfolio for the Post-COVID Era
Presenter: Dan diBartolomeo
September 30, 2021
Abstract
The COVID-19 pandemic has focused investor attention on the concept of resilient portfolios that can reasonably be expected to perform well through future periods of market stress.
We will begin with a review of two published papers that provide the theoretical basis and empirical data for the current study. The first paper, diBartolomeo and Kantos (Journal of Asset Management, 2020) restates the past thirty years of style factor returns conditional on the assumption that investors are explicitly concerned about the risk of large effects (e.g. pandemics). The second study diBartolomeo (Journal of Performance Measurement, 2020) focuses on four specific types of events (war, pandemic, climate change, and pervasive corruption).
We will then use this information to optimize a typical US equity portfolio that we expect to provide superior risk-adjusted performance in the event of market stress.
We will conclude with illustrations of “optimistic” and “pessimistic” portfolio construction for investors with different views of the frequency and severity of future stress effects.
Northfield WealthBalancer: A Use Case
Presenter: Mike Knezevich
September 16, 2021
Abstract
Northfield has made major strides in the private wealth space. From determining sound investment goals specific to an individual’s needs to implementing those goals into a tax efficient optimal portfolio personalized to the individual investor.
With the implementation of our services, Northfield has turned what has traditionally been perceived as a cost center into a profit center by automating customized portfolio construction en masse. Not only does each investor receive a personalized optimal portfolio, but automation frees up the advisor to provide a higher level of customer service to build their business.
Most advisory and robo-advisory systems focus on the front end with very little emphasis on creating a rigorous investment plan. WealthBalancer provides the in-depth theory rich back end to construct the most appropriate asset allocation by:
· Determining the level of risk an investor can afford and their level of comfort with risk.
· Customizing the universe to asset classes the investment manager deems appropriate for the investor. Depending on the investment firm’s objective the universe may include asset class representation from which to build a portfolio, mutual funds available from the investment firm and/or single securities.
· Allowing the investment firm to implement their own Capital Market Assumptions and Expected Returns. Adding value from the firm’s insights.
· Differentiating investments in either taxable or non-taxable to best suit the tax treatment of the individual asset for the investor’s tax circumstances.
· Projecting asset allocations over time based on the investor’s goals.
Based on my June 2021 Newsletter article available at https://www.northinfo.com/docs/tech0621.pdf, this product workshop illustrates and extends upon WealthBalancer’s functionality using a hypothetical example.
I use the demo GUI for purposes of illustration during the workshop. It is important to note that a flexible WealthBalancer API is available for users to develop their own white labeled front end that would better represent their firm and client base.
Returns from Cost Centers: Quantifying Asset Owner Alpha from Better Risk Management and Trade Execution
Presenter: Dan diBartolomeo
August 31, 2021
Abstract
Many large asset owners manage internally as if they were really an asset manager with just one client (themselves). This blurring of roles masks a key difference in the economics of the situation. For agent managers, the cost of high-quality risk management efforts are borne by the manager, while the economic benefit of better risk control accrues to the investor client. The same separation of costs and benefits applies to the cost of improving trade executions. Managers bear the operational costs while clients get almost all the benefit.
However, for a large asset owner with internal management both the costs and benefits accrue fully to a single entity making for a very different situation.
In this presentation, we will provide two analytical models that quantify the return improvement arising from better risk control and lower trading costs. We assert that a “best of breed” risk management effort will pay for itself hundreds of times over, while improving trading operations yields lower but still triple digit “returns on investment.”
Advanced Risk Decomposition
Presenter: Steve Dyer
August 17, 2021
Abstract
This user workshop will cover two ways of digging deeper into the risk of your portfolio on a more granular basis than the standard risk decomposition reporting. To explore risk contributions from individual positions, we will first look at the often ignored implied alpha metric, which is a useful metric for all managers to understand why they are holding the weights they do.
From there, we will drill down one more level to show how to calculate each position’s contribution to risk on a factor-by-factor basis. This allows the user to decompose the risk of the portfolio by position, and more importantly, decompose the risk coming from subsets of positions: by manager, by asset class, by sector, by percentile, or by any other grouping.
We hope that users will leave this workshop with a deeper understanding of the risk reporting offered by Northfield, as well as some new ideas about how to communicate that information across your organization.
Estimation of Corporate Bond Credit Ratings
Presenter: Dan diBartolomeo
July 29, 2021
Abstract
In the more than twenty years since Northfield’s adoption of the “Everything, Everywhere” concept, we have built internal models to fill gaps in the available information set required to do a credible job of multi-asset class risk assessment. Traditional credit ratings are issued by rating agencies in structure where the borrower rather than the lender is paying for the credit rating. This incentive structure creates inherent conflicts of interest that have led to serious biases in ratings, as contributed greatly to the Global Financial Crisis (2007-2010). This issue was addressed in our 2011 essay, Five Easy Steps to Fixing the Rating Agencies (https://www.northinfo.com/documents/489.pdf).
Secondly, while formal ratings are available for the majority of traded assets when measured by market value more than two thirds of the 90,000 corporate entities covered in our global risk models do not have a published rating. Many ratings issued by small, local rating agencies have proven to be extremely unreliable (i.e. a default by a AAA rated entity).
Finally, rating agencies are often slow to respond to changes in conditions at the macro level or a particular entity. For example, the average time between rating changes at major rating agencies is more than three years.
This webinar will be devoted to demonstrating the analytical process by which Northfield can now provide ratings for all 90,000 corporate entities. The process is an extension of Merton (Journal of Finance, 1974) as proposed by diBartolomeo (Journal of Investing, 2010). The current process also includes a Bayesian process to capture industry level differences in accounting procedures (e.g. airlines). We will present historical data showing that the 2010 version of our model significantly outperformed traditional ratings during the GFC, so we may reasonably expect that the newly improved process will be even better.
The presentation will close with a preview of future planned improvements including the overlay of a non-parametric model, similar to those of Srinivasan and Bolster (1990), and Bolster, Johnson and Srinivasan (1995) which will allow for coverage of private loans and other assets associated with traditional banks.
Tax-Aware Optimized Back-Testing Using the Northfield Optimizer and Risk Models
Presenter: Luke Smith
July 13, 2021
Abstract
In order to illustrate the many challenges and intricacies of simulating the multi-period evolution of a tax-aware portfolio, we will present a case study pertaining to a young investor with a modest income, a small starting portfolio, and a long investment horizon.
With the advent of commission free trading and fractional shares in recent years, tax-sensitive investing has migrated from the exclusive realm of high net-worth investors and their well-paid advisors to investment novices with a few thousand dollars and their low-fee robo-advisors. The actual amount of “tax-alpha” that is achievable is a frequently asked question. We will demonstrate that the answer varies widely since it is highly dependent on the investor’s income and portfolio size, as well as the performance of the benchmark over the relevant period.
In order to calculate realistic after-tax returns, it is necessary to decompose stock returns into price changes (capital gains) and dividends, and to accurately simulate and apply the complex tax regulations that relate to both types of investment income. These rules include the netting of short-term and long-term realized gains and losses, the annual $3,000 capital loss allowance, loss carryovers, wash sale rules, taxes related to short positions, qualified vs. unqualified dividends, and the treatment of spin-offs and acquisitions for cash and/or stock. The applicable regulations and current tax rates will be reviewed during the presentation.
Highlights will include:
· The distinction between pre-liquidation after-tax return and post-liquidation after-tax return.
· An analysis of the tax benefit of the $3,000 annual loss allowance for investors with various portfolio sizes and marginal income tax rates.
· A discussion of the “outside gains” assumption and its inclusion in after-tax returns.
· A simple approach for including taxes on dividends in the objective function.
· A theoretical upper limit on tax alpha given portfolio size, tax rates, and benchmark performance.
Data Sufficiency for Risk Management for Financial Institutions
Presenter: Dan diBartolomeo
June 29, 2021
Abstract
In this presentation, we will be presenting a pragmatic framework by which various types of financial institutions can judge the completeness and sufficiency of their asset data. We will also provide a preview of a partnership among Northfield and other firms that will allow our clients to ensure complete and sufficient data on their asset risk exposures is available conveniently at modest cost.
Simply put, many financial institutions do not know everything they own (or have lent against) or do know in such poor detail that competent risk management processes needed for today’s complex markets are impossible. It would seem common sense that “if you don’t know what it is, don’t own it” would be the operative rule for institutional investors but such standards are rarely observed.
Unrelated to the traditional concept of market declines, numerous failures of basic risk management have been at the center of financial news in recent months. Examples include the staggering $11 Billion in losses at five major firms who financed the Archegos fund, to the $3 Billion collapse of institutional investor funds associated with Greensill, and the $2 Billion in losses taken by a single Canadian pension fund in variance swaps during the early pandemic period. In a previous essay, we described the folly of an anonymous but real financial institution reducing the accuracy of their risk management efforts in order to achieve small savings in operating costs, The Economic Cost of “Good Enough” Risk Management for Asset Owners, https://www.northinfo.com/documents/850.pdf. The essay was prescient in that this specific institution sustained multibillion-dollar losses in the matters above.
In some ways it seems little has changed from days of the Global Financial Crisis and the Madoff fraud. In response to such events, regulators in countries from Australia, India and the USA are tightening risk standards and reporting for banks, asset managers, and asset owners. For risk management processes to have any chance of success, an institution must have a 100% complete asset master (both public and private) that is organized for investment purposes as distinct from the format of a traditional financial statement. Secondly, the asset master must contain descriptive data on each line item of exposure that is sufficiently granular to allow risk analytics to operate, including estimated current market values for illiquid positions.
Cryptocurrency Risk
Presenter: Dan diBartolomeo
June 15, 2021
Abstract
The market value of all cryptocurrency now exceeds 1% of all traded wealth. In recent months there have been massive swings (both up and down) in the values of cryptocurrencies like Bitcoin and Ethereum. Some financial institutions such as Goldman Sachs are setting up trading desks to participate in these assets, while other institutions such as HSBC has explicitly stated that they do not support any participation by their clients. Irrespective of their intrinsic and extrinsic value, we expect that such items will be turning up in client portfolios from time to time. As such, we will be covering the more prominent cryptocurrencies as a currency in future risk model releases.
During this presentation, we will be explaining a number of key building blocks for understanding how to evaluate the risk of cryptocurrencies and what magnitude of return expectations would justify those risks for a typical investor.
We will show how our basic methodology uses forward-looking information in addition to historical data, as well as several of the statistical nuances of the estimation process. Factor loadings for cryptocurrencies will be a combination of certain fiat currencies (e.g. Swiss Franc) and commodities (e.g. gold) in a process similar to our updated analysis of commodities (see Northfield News March 2021, https://northinfo.com/documents/990.pdf.
The presentation will conclude with an empirical comparison of our estimates for Bitcoin volatility to the BVI (a thinly traded contract on Bitcoin volatility, similar in concept to VIX).
The Real Economics of ESG
Presenter: Dan diBartolomeo
May 27, 2021
Abstract
While the concept of “ESG” investing has become vastly more popular in recent years with aggregate AUM of explicit ESG funds approaching $1 Trillion, the evidence on the financial benefit of such practices to equity investors is mixed. E, S and G strategies all have different levels of “free riding” as everyone in the world might benefit from a slowing of climate change, while the benefit of better governance of a given company accrues to a far narrower set of stakeholders. Although numerous academic studies have been conducted, there is not a clear answer as to whether investment strategies built around “environmental, social, and governance” or the related concepts of “social responsibility investing” or “sustainability” actually produce different investment outcomes.
Many studies have reported differential return outcomes (both positive and negative) but have been unable to positively associate the differences with ESG/SRI effects when we control for traditional equity attributes for time periods prior to the recent newfound popularity of such strategies. If we observe only recent data, we may attribute excess returns to ESG as a sort of self-fulfilling prophecy. Equity holders might benefit over short time horizons from a company reducing expenses related to environmental safeguards or good treatment of their labor force even if such “corner cutting” increases the likelihood of a “large” negative event in the long run (e.g. an oil spill or labor strike).
In this presentation, we will resolve this question based on a consistent finding across studies of ESG effects on corporate bonds. For bond investors there is no financial “upside” if companies cut corners that increase long horizon risks. Multiple studies show that regular (i.e. not special “green bonds”) corporate bonds of companies with high ESG scores trade at lower yield spreads. To accept lower yields, bond investors must believe the bonds to have less credit risk over their lifetime. It has also been reported that corporate bond mutual funds with higher ESG scores have outperformed on a risk adjusted basis. Since the yields are generally lower, the superior risk adjusted performance must arise as the result of lower risks.
Using the method of “contingent claims analysis” from Merton (Journal of Finance, 1974) and diBartolomeo (Journal of Investing, 2010) we will estimate the magnitude of the risk reduction associated with ESG attributes. We will conclude by showing that these results are consistent with our previously reported findings on long-lived “sustainable” companies, Northfield News December 2018, https://www.northinfo.com/documents/848.pdf.
If It Walks Like a SPAC, and Quacks Like a SPAC, Is It Private Equity?
Presenter: Emilian Belev
April 27, 2021
Abstract
The past year gave rise to a record number of Special Purpose Acquisition Companies (SPAC) entering the public equity market. Often times these have been heralded as liquid private equity - offering investors the perceived upside of private company investments, but also the liquidity of public stocks. This view has potentially been reinforced by a number of traditional private equity management firms joining the proliferation of SPACs.
In this presentation, we will analyze the key similarities and differences between a SPAC and a typical Limited Partnership private equity fund and paint the nuances of how the risk-return characteristics between the two types of investment vehicles would differ.
Then, in turn, we will address the question how much SPACs can be considered a replacement of traditional private equity fund investments.
Representation of Illiquid Asset Exposures in the Liquid Markets
Presenter: Mike Knezevich
April 13, 2021
Presentation Abstract
Northfield uses a sophisticated bottom-up approach to model illiquid real estate investments without relying on liquid securities. In the March 2021 Northfield Newsletter, https://www.northinfo.com/docs/tech0321.pdf, I wrote that situations may arise in which investors want to represent all or part of such an illiquid investment with liquid securities. Using an investment in an NYC office building modeled with the bottom-up approach, I demonstrated how:
· To create a synthetic asset mimicking the factor exposures of the office building. Such synthetics are desired when an investment decision has been made to buy or sell a real property with a relatively long closing process. To fully implement the investment decision prior the transaction being completed, the investor may choose to create the synthetic to include in their portfolio.
· The current owner of a real estate investment who wants to hedge part of the real estate asset which they believe will be riskier or under perform. Since it is impossible to sell just the portion of the building they no longer want to be invested, liquid securities can be added to the building portfolio to mitigate the undesired exposure(s).
In this workshop, using the Northfield multi-asset class risk model (Everything Everywhere or EE) and Northfield’s newest hosted application (Nexus), We will demonstrate two optimizations. We will begin with a risk decomposition of the office building to understands how illiquid assets are modeled in EE. Next, we will construct a portfolio of liquid securities to fully replicate the factor exposures of the office building. Finally, we will construct a hedge against a portion of the office building’s exposure.
GameStop, Variance Swaps, and Related Failures of Hedge Fund Risk Management
Presenter: Dan diBartolomeo
March 25, 2021
Abstract
Recently there have been some highly publicized failures of hedge fund risk management. These events resulted in very large losses for hedge fund investors. Hedge fund activity at multiple firms (including Fidelity) has been discontinued.
In this presentation, we will describe persistent weaknesses in the way that hedge funds have managed risk in their portfolios and illustrate how such failures could have been avoided. The common theme of these values is the improper assessment of how particular investments create higher moments in the distribution of fund returns.
The first thematic failure connecting these events is an implicit assumption that liquidity is always more than sufficient and therefore risk management practices can assume that hedging relationships (e.g. a long/short portfolio) will be rebalanced continuously. While the assumption of continuous rebalancing leads to elegant closed-form math for pricing many derivatives, reliance on this assumption leads to biased estimation of risk. For example, if you can only rebalance a market neutral portfolio periodically, the maximum loss is 100% of the value of the long side, but the maximum loss on the short side is unbounded. This means that distribution of fund returns will have negative skew and positive kurtosis.
The second failure is the intentional adoption of strategies (e.g. selling variance swaps) that have negative skew so that the fund returns on average are higher but where extreme losses are far more frequent than would be expected under a normal distribution (although still relatively rare).
The third issue is the persistent failure to account for the implications of boundary conditions associated with observable market data. For example, the implied volatility of various GME options reached over 500% annualized at certain recent points. The annualized variance of 250,000%2 means that the expectation of the geometric mean return for GME would be about negative 5% trading per day.
Using data from the Northfield Short Horizon and Near-Horizon risk models, we will illustrate how these situations have been properly addressed for our client hedge funds and their investors.
Strategy Design and the Fallacies of Breadth
Presenter: Leigh Sneddon
March 9, 2021
Abstract
High return correlations between assets have widely and plausibly been said to detract from active performance. The concept of Breadth supports this view. Buckle, however, reported that these correlations improve active performance.
Based on three approaches, we report that Buckle is correct. First, we introduce a model of active investing that includes return correlations between assets. Design formulae give the time averages of active risk and return and other portfolio characteristics. Next, we show that Monte Carlo simulations confirm the conclusions of the model. Finally, we provide the investment intuition behind the conclusions.
All else equal, active managers should be more aggressive when return correlations are expected to be high, not low. A related widely held belief is that return forecasts with lower correlations between assets provide better performance. The model also includes these correlations. The same approaches show that this consensus too is incorrect: all else equal, correlations of return forecasts between assets improve performance. The design formulae can be used for operational strategy design aimed at performance improvements.
As examples, we use it to exploit differences in predictive power between sectors and to efficiently incorporate Environmental, Social and Governance controls into active security selection.
Fund Commitment Planning for Target Private Asset Class Allocations
Presenter: Emilian Belev
February 25, 2021
Abstract
In a shifting landscape of allocations of asset owners, the ability to implement a trajectory to a particular allocation target for each asset class is key. One of the challenges that private asset classes represent is that they often come in the form of Limited Partnership funds where capital is committed once but is called over a number of years.
This means that there is a disconnect between the amount committed in each of the coming years, the amount that will be called in during the same periods, and the evolution of net asset value of the private asset class based on commitments made prior to each point in time. The investor can certainly make an outsized level of commitments to private funds to assure a minimum level of allocation, but that carries the obvious risk to overcommit. Over-commitment, in addition to overshooting the allocation target, often means taking liquid resources away from tactical uses and placing them into segregated facilities to meet a potential slew of capital calls.
In this presentation we discuss a framework whereby the investor can gauge the relationship between these three variables – commitments, capital calls, and NAV. This relationship provides the ability to deal with all allocation challenges simultaneously, plan optimal liquidity in view of capital call considerations, and implement a path to an allocation target for the private asset classes. In doing so, we address both the uncertainty of net asset value of the specific asset class, as well as the uncertainty of NAV for rest of asset classes in the investor’s portfolio.
Modeling Fixed Income Liquidity and Trading Costs
Presenter: Dan diBartolomeo
February 11, 2021
Abstract
As a financial market still dominated by dealers rather than centralized exchanges, the accurate assessment of liquidity and trading costs for fixed income securities has lagged far behind other equities, currencies, and many derivatives.
While transparency has improved somewhat through online data services like TRACE and ZEN, the ex-ante estimation of fixed income liquidity remains problematic. In addition, the rapid rise in the popularity of ETFs as the line has blurred between the routine liquidity of ETFs and the limitations on liquidity of the underlying constituent securities.
In this presentation, we will first illustrate how liquidity metrics required by regulations (SEC and MIFID II) are likely to make this problem worse rather than better. We will then present three alterative procedures for addressing liquidity concerns in fixed income portfolio management, including a fixed income extension to the existing Northfield transaction cost model.
Fund Commitment Planning for Target Private Asset Class Allocation
Presenter: Emilian Belev
January 28, 2021
Abstract
In a shifting landscape of allocations of asset owners, the ability to implement a trajectory to a particular allocation target for each asset class is key. One of the challenges that private asset classes represent is that they often come in the form of Limited Partnership funds where capital is committed once but is called over a number of years.
This means that there is a disconnect between the amount committed in each of the coming years, the amount that will be called in during the same periods, and the evolution of net asset value of the private asset class based on commitments made prior to each point in time. The investor can certainly make an outsized level of commitments to private funds to assure a minimum level of allocation, but that carries the obvious risk to overcommit. Over-commitment, in addition to overshooting the allocation target, often means taking liquid resources away from tactical uses and placing them into segregated facilities to meet a potential slew of capital calls.
In this presentation we discuss a framework whereby the investor can gauge the relationship between these three variables – commitments, capital calls, and NAV. This relationship provides the ability to deal with all allocation challenges simultaneously, plan optimal liquidity in view of capital call considerations, and implement a path to an allocation target for the private asset classes. In doing so, we address both the uncertainty of net asset value of the specific asset class, as well as the uncertainty of NAV for rest of asset classes in the investor’s portfolio.
Risk Decomposition Under Parameter Uncertainty and Price Movement
Presenter: Anish Shah
January 14, 2021
Abstract
Portfolio managers are well acquainted with risk decomposition, a canonical tool for identifying and estimating sources of risks in a portfolio. As risk always exists in the future, a proper understanding of risk requires we consider our level of statistical confidence in the assessment and decomposition. The literature has long recognized the importance of addressing statistical estimation error (Ledoit and Wolf, 2004) in historical covariance matrices. Newer non-linear methods can also be applied to factor models of risk (Ledoit and Wolf, 2020).
However, an important element of the problem has been ignored: the extent to which we expect future price movements to change portfolio asset weights during the finite future time interval during which the risk assessment is presumed to hold true. Ignoring changes in expected risk arising from security price movements is similar to the situation of option hedging where textbook formulae for parameters such as "delta" unrealistically assume that hedging relationships are rebalanced continuously.
This presentation will provide an original analytical approach to factor risk decomposition that is inclusive of both estimation error and the expected slippage in hedging relationships amongst portfolios assets that will arise from asset price movements. Much like a medical X-ray, this enhanced form of risk decomposition reporting provides a much richer and more stable understanding of portfolio risk.
We look forward to implementing this advance in future Northfield systems.
Advances in Portfolio Customization
Presenter: Dan diBartolomeo
December 22, 2020
Abstract
Recently there have been multiple high-profile acquisitions of asset management firms (e.g. Parametric, Aperio) that are distinguished by their relative proficiency at customizing portfolios to investor needs and preferences beyond the simple return and risk. Driven by the pressures of low fee passive management, it would appear that the industry has reached the “tipping point” where portfolio mass customization will be a routine requirement for success, as it has been for decades in other industries (e.g. automobiles where buyers can order a vehicle to fit to their exact preferences of model, color, and optional equipment). Typical areas of customization include tax sensitive management of portfolios, inclusion of non-financial concerns of investors (e.g. climate), and explicit recognition of investor’s financial goals such as retirement income.
While Northfield has been a leader in portfolio customization since the introduction of tax-sensitive optimization in 1996, only a handful of top-tier asset management firms have taken significant advantage of the analytical techniques that were already available a quarter-century ago. In the intervening period, we have continued to innovate in both the analytical aspects of customization (e.g. non-parametric investor preference functions) and in the operational platforms that now allow asset managers to efficiently scale their customization efforts to hundreds of thousands of portfolios. We also contributed to a book published by CFA Research institute to define best practices for the wealth management industry (diBartolomeo, Horvitz and Wilcox, 2006). Unfortunately, many vendors have jumped on the customization bandwagon with processes so unsophisticated as to be potentially detrimental to investors.
In this presentation, we will provide the Northfield framework for mass customization and describe the wide array of operational processes that have evolved that can allow all wealth management organizations to cost-effectively offer institutional quality customized portfolios at large scale.
The final portion of the presentation will be devoted to introducing several analytical innovations that will further enhance investor outcomes, such as the use of machine learning techniques to create written narratives for investors explaining portfolio investment policies and transactions.
Portfolio Construction Under Economic Scenarios In Action
Presenter: Mike Knezevich
December 8, 2020
Abstract
In September 2019, Northfield hosted a webinar entitled “Portfolio Construction Under Economic Scenarios.” The basis of the webinar was to discuss how investment organizations can successfully implement economic scenarios into an optimization.
Investment organizations spend a lot of resources forecasting future economic conditions. Implementing these economic forecasts into constructing a portfolio, however, has proven to be a difficult task. During the September 2019 webinar, Northfield President Dan diBartolomeo discussed using Northfield’s updated functionality and multi-asset class risk model to convert economic forecasts directly into forecasts of returns, volatility, skew and kurtosis for any individual security or any set of securities making up an asset class (i.e. a index portfolio). Such economic forecasts can fall within the range of very simple to being as complex as the data permits. This new functionally allows users to construct an investment portfolio or asset allocation under various economic circumstances.
During the previous webinar Dan mentioned this functionality would be automated and available in Northfield’s PRISM system.
In this online workshop we will demonstrate how to set up and use an economic scenario analysis in PRISM. We will then illustrate how easily and seamlessly the scenario is integrated into the Northfield Open Optimizer within PRISM. We will begin by exploring the impact of a simple one economic factor scenario on the construction of our portfolio. Then we will explore the impact of a more complex economic scenario.
Constructing Private Equity Benchmarks
Presenter: Emilian Belev
November 24, 2020
Abstract
With the evolution of private assets from an exotic experiment to sought-after driver of yield and return in institutional portfolios, serious challenges emerged to appropriately review and forecast benchmark relative performance. The main difficulty relates to the fact that unlike publicly traded investment the values and returns reported by private asset investment managers are very subjective, making it difficult to perform the usual benchmark construction process where one can rely on a particular investment or a combination of investments to be an appropriate and unbiased gauge for the relative performance of another investment.
We will make an overview of the attempts made by the industry to operate in the face of this challenge. Consequently, we will describe four approaches that, with some modifications, overcome the shortfalls of pre-existing methodology, and have distinct applications in the institutional investment process.
Those methods can be categorized as the private-to-private returns, private-to-public-plus-margin returns, public-discounted-paid-in multiples, and liability-relative benchmarking.
RAMP - A Risk Reporting Service for Asset Owners and Managers
Presenter: Richard Dawson
November 12, 2021
Abstract
This presentation will take an in depth look at the reporting capabilities of Northfield's RAMP service. After a brief overview of the system, we will demonstrate the recent enhancements to the service, showcasing the full power of the reporting system and the depth of analytics available. RAMP provides time series reporting, stress testing and scenario analysis along with fully customizable risk breakdowns.
Built around Northfield's Everything Everywhere model, RAMP's analytics cover all asset classes, including illiquids such as real estate, complex derivatives and other alternative assets.
We'll focus on a particular set of reports produced for a multi asset class, multi national portfolio.
Simplified Investment Performance Evaluation (SIPE)
Presenter: Dan diBartolomeo
October 29, 2020
Abstract
In this webinar we will present a new measure for risk-adjusted return (SIPE) which can be used for universal evaluation of investment fund performance. SIPE has numerous advantages over traditional measures such as excess returns, Sharpe ratio, information ratio, and realized mean-variance utility.
SIPE is equally applicable to absolute or benchmark relative returns, or any blend of investor preferences that incorporate both. SIPE is also self-customizing in that the scaling of the penalty for risk undertaken is inferred from the current portfolio. This makes SIPE both appropriate for all asset classes and across the full spectrum of conservative to aggressive investors. The structure of SIPE also explicitly captures an often neglected property of active management.
All active managers (and their investors) must rationally expect to outperform fair benchmarks. If some investors achieve outperformance other investors must underperform. As such, a significant portion of active investors must always be wrong in their expectation of outperformance.
SaaS and Risk Model Data
Presenter: Daniel Mostovoy
October 15, 2020
Abstract
One of the key problems in portfolio analytics is finding risk model records for the assets in a user’s portfolio. Often the user has a mix of asset identifier types or perhaps the risk model doesn’t cover all the assets in the user’s portfolio. Or maybe the user has real estate, private equity, or other “alternative" positions that need to be represented via some customized process.
This talk addresses how the Northfield Analytics Service Platform, (SaaS.northinfo.com) tackles these issues, through automated id cross referencing, proxying missing assets and private risk model data sets.
Other topics we will cover are - access to risk model data via website, JAVA and RESTful webservice API & the convenience of keeping risk model data offsite.
Private Real Estate Cash Flows Income: Alternatives Move Into the Mainstream
Presenter: Emilian Belev
September 29, 2020
Abstract:
The yield curve dynamics of the last decade has relentlessly reshaped the role of traditional investment grade bonds in the portfolios of multi-asset class investors. The pandemic of 2020 has been the sprint at the end of the marathon in the same direction.
Due to a scarcity of traditional low-risk assets providing sufficient income capacity to meet future mandated targets, investors cognizant of future liabilities and expenditures have an increasingly hard time to maintain the probabilities of shortfall in the medium- and long-run at tolerable levels. For this reason, paradoxically, such investors have made significant advances of including asset classes like private equity, private debt, and private infrastructure and real estate, as they engage further on risk dimension of the Prospect Theory curve.
Real Estate being a prominent example of an income producing alternative asset class is an excellent candidate to illustrate the risk-return impact of this investment trend. For long we have maintained and found strong correlations between directly owned real estate and traditional fixed income instruments. Over time we have evolved the capability to evaluate the specific future income producing capacity of the real estate asset class and the cost of uncertainty with which this income is associated, both in periodic and cumulative performance sense. This type of analysis is the cornerstone to address the objective of minimizing the probability of shortfalls and the impact of real estate in a diversified asset class portfolio and as such is the focus of both academic and industry practitioners.
Analyzing Historical Risk in Nexus
Presenter: Lalitha Raman
September 8, 2020
Abstract:
Nexus is Northfield’s new online platform to access best-in-class Northfield analytics and services, providing a window into our entire suite of models across all time horizons.
Nexus offers a fresh new interface over the web with added features to analyze factor-based portfolio risk, performance attribution and optimization.
This workshop will build upon our July 2020 “Introducing Nexus” workshop by demonstrating the risk module service’s historical processing feature.
We will showcase Nexus reporting capabilities as we analyze historical risk over this unprecedented period in recent history. We invite you to see what our models tell us about risk forecast through time using our long term and near horizon models for a global portfolio over the course of 2020.
A Comparison of Conditional and Regime Switching Methods for Equity Risk Models
Presenter: Dan diBartolomeo
August 27, 2020
Abstract:
The recent volatility in global equity markets driven by the COVID-19 pandemic has reminded investors that both perceived and expected risk levels change from time to time as events unfold. Such circumstances beg the question as to how to best structure equity factor risk models to correctly reflect these changes.
In this presentation we will compare regime switching and conditional formulations for equity risk models. Since 1997, Northfield has pioneered the fully conditional factor risk model. In such models, expectations of future volatility are conditional on contemporaneously observable information which is explicitly forward looking. Northfield uses three kinds of conditioning information (option implied volatility, state variables [VIX and credit spreads], and text analytics). The most widely used is the content of financial news text, which is evaluated in our Risk Systems That Read® process.
An entirely different framework that has been widely applied in the area of dynamic asset allocation is a “regime switching” model, wherein we believe that at each moment in time, global markets exist in one of two or more identifiable different states. Each state has its own probability distribution for returns, and covariance (in effect, a separate model). Adherence to such concepts has led to the popular idea of “risk on, risk off” wherein investors switch their portfolios between what they believe to be appropriate allocations for the various distinct states. The use of these concepts has been largely motivated by a series of four closely related research publications that began with Chow, Jacquier, Kritzman, and Lowry (Financial Analyst Journal, 1999).
We find that at the level of analyzing portfolios of individual securities, conditional methods are superior to regime switching equity models on both practical and theoretical grounds, as well as the results of four different empirical analyses. Limited support is found for a third approach, ARCH/GARCH type models.
Household Optimization for Private Wealth and Family Offices
Presenter: Steve Dyer
August 11, 2020
Abstract:
Please join this online workshop for a demonstration of the capabilities of Northfield's new householding optimization feature, which allows managers to perform joint optimizations of multiple accounts for multiple members of a household without sacrificing customization or specificity.
The benefit of using the household optimization is that managers can balance the preferences and idiosyncrasies of every member of the household while gaining synergies across accounts and keeping the whole household on track.
Some of the benefits of household optimization include:
This functionality is available in the Northfield Open Optimizer now, so join this workshop to learn how you can simplify your operations while increasing the sophistication of service you can provide to clients.
Co-Investing for Limited Partners: Another Risk vs. Return Tradeoff
Presenter: Emilian Belev
July 30, 2020
Abstract:
As asset owners practices in private equity investment matured over the course of the last decade, a number of them found co-investment an attractive alternative to traditional LP investing.
The attraction of this type of equity participation is minimal management fee expenses while still maintaining a passive role in sourcing the deal flow and the associated costs. The less talked about aspect of this type of investing is the requirement of large stake participation and the reduced capacity for diversification that comes with it.
This talk will focus on the key metrics that allow the LPs to find the right balance between the additional upside potential and the higher level of investment risk, where "beta" risk is not the primary concern. By extension, we will discuss some implications on the transparency of limited partnership agreements with regards to co-investment opportunities, as well as the recent regulatory focus on the consistency with which these clauses are implemented and the repercussions both on the GP and LP side, as an additional source of risk.
Introducing Nexus - Northfield's Integrated Analytics Platform
Presenter: Ghazanfer Baig
July 16, 2020
Abstract:
In this workshop we will introduce Nexus, a new online platform to access best-in-class Northfield analytics and services. Nexus offers a fresh new interface over the web with added features to analyze factor-based portfolio risk, performance attribution and optimization.
The workshop will:
Join us as we unveil Nexus and be among the first to gain access!
What the Coronavirus Pandemic Taught Us About Factor Returns: Real Alpha is Better than "Smart Beta"
Presenter: Dan diBartolomeo
June 25, 2020
Abstract:
The coronavirus pandemic has taught us a valuable lesson in terms of how factor returns and "alpha" should be interpreted.
In this presentation, we will discuss the more than thirty years of factor return history from Northfield's US Fundamental Model, after slightly modifying the structure to account for the presence of "rare, extreme events" (e.g. the pandemic). The resultant model is consistent with the Black (1972) version of CAPM where the zero-beta asset is has no correlation with the market portfolio but may have non-zero volatility. It is also consistent with the work of Barro (2005) and Gabaix (2009) on rare events.
The results confirm the existing consensus of positive alpha for "value" and "momentum" effects. More interesting, we demonstrate both a statistically significant positive return to CAPM beta as a risk factor and a significant negative return to absolute volatility. The only way both can be true is a negative return to idiosyncratic risk which we show is related to corporate bankruptcy. different
Constructing Specialized Portfolios
Presenter: Steve Dyer
June 11, 2020
Abstract:
In this webinar, you will be guided through how to construct four different specialized portfolio types that have been the subject of interest in different market environments: minimum variance, maximum diversification, risk parity or "all weather" and factor-mimicking portfolios.
Along the way, we will be discussing different frameworks to think about portfolio construction and pointing out optimization best practices and lesser-known features and functions in the Northfield Open Optimizer.
This webinar is for all users of the Northfield Open Optimizer who want to gain a deeper practical knowledge of the product's capabilities
The Four Horsemen of the Investment Apocalypse: Pandemic, War, Corruption and Climate Change
Presenter: Dan diBartolomeo
May 26, 2020
Abstract:
While the world continues to struggle with the impact of the coronavirus pandemic, investors need to remain cognizant that other thematic effects can be vastly influential on investment outcomes in both the near term and the long term. Of our four themes, pandemic and war are sufficiently violent and episodic (hopefully) as to dominate economic news and investor thinking in the moment. The other two themes, corruption and climate change affect investment outcomes in a slow acting, persistent fashion in ways that are often too subtle to be captured by a conventional risk system. Much like the fable of the "tortoise and the hare," it is unclear which of these thematic effects will ultimately win the race for greatest influence on global investment outcomes. As such, global investors must be prepared to address all four.
We will begin with a brief update of our two recent research articles on the impact of the ongoing coronavirus pandemic. The next portion of the presentation will be devoted to our previous research on the financial market impacts of war and societal corruption (diBartolomeo and Hoffman 2015) and a recent Northfield article on methods for addressing climate change in asset portfolios.
The final portion of the webinar will demonstrate how three analytical techniques arising from these research efforts can be applied to insulate investor portfolios from all four thematic effects.
Asset Allocation Dislocation: Extent, Impact, Solutions
Presenter: Emilian Belev
April 30, 2020
Abstract:
This presentation will focus on the relationship between market and book NAVs across public and private asset classes and its effects on portfolio rebalancing.
The most recent pandemic-driven market slump exposed the stark divide between the views and practices on valuations of private and public assets. The discrepancy poses a real challenge for asset owners with policy target allocations.
We will explore the origins of the problem, the extent to which it is economically founded or accounting practice driven, and whether the solution involves selling illiquid assets, curtailing commitment programs, or taking a fresh view at private asset risk aware valuations.
Estimating Investor's Return-Volatility Trade-Off: The Answer is Always Six
Presenter: Dan diBartolomeo
March 26, 2020
Abstract:
A constant but vexing question among investors is "How much incremental return must I expect to justify a particular increase in the risk of my portfolio?".
In this webinar, we will derive a "rule of thumb" that efficiently describes the optimal return-volatility trade-off parameter for a wide range of practical cases. Since the seminal paper of Levy and Markowitz (1979, Journal of Finance), the most popular way to describe investor objectives has been the mean-variance utility function. The alternative method of Sharpe Ratio has also been shown to be equivalent to mean-variance under certain circumstances as described by DeGroot and Plantinga (2001, Journal of Performance Measurement).
However, many investors find statistical variance an unintuitive measure and so prefer to think of the trade-off between return and risk with the unit of risk being standard deviation or a scalar of standard deviation such as VaR or CVaR. Most investors are simply unable to numerically express their mean variance trade-off parameter ("lambda" or it's percentage reciprocal which Northfield terms "RAP") with any confidence. While marginal properties of any particular mean-variance objective can be mapped to mean-standard deviation using the "chain rule" of calculus, a very different and pragmatic approach to this question can be derived from the methodology in Wilcox (Journal of Portfolio Management, 2003).
In this process, we assume that the investor has chosen their current level of portfolio risk so as to bound a "worst case scenario." From this boundary condition we can infer the mean variance risk tolerance as a scalar function of the current portfolio expected return and volatility. Algebraic simplification leads to the conclusion that for a wide range of situations and asset allocations, the optimal trade-off parameter between incremental expected return and incremental volatility is typically about one sixth. For example, a 2% increase in portfolio volatility is reasonably justified by a .33% (2/6) increase in expected return. The method works identically across absolute risk, tracking error or any desired blend of the two risk measures
Multi-Period Correlations across Public and Private Asset Classes
Presenter: Emilian Belev
February 25, 2020
Abstract:
In this webinar we will address one challenge of long-term investors, namely the ability to gauge correlations and volatilities of asset classes over multi-year investment horizons.
Implicitly the focus of most commercial risk models in existence is the performance, volatility, and correlation among assets over one period, whether this is a day, month, or a year. There have been a number of formulaic and simulation-based extensions that provide some insight as to the long-term volatility of investment assets that transform the one period risk model estimates into multi-period ones. However, there has been much more limited published work and commercial solutions that address the multi-period correlations among assets.
We will demonstrate how using Northfield and partner technology generates robust distributional performance projections over any time horizon, we can utilize a conventional risk model to build a correlation matrix of assets and asset classes over a number of periods of our choosing.
Just like the implications of multi-period volatility, our findings will provide useful insights as to the risk-return profile of the multi-asset class portfolio of a long-term investor.
A Heuristic Approach for Delta Hedging in Discrete Time
Presenter: Dan diBartolomeo
January 30, 2020
Abstract:
One of the most basic concepts in modern finance is that of "delta hedging" of an option. It can be shown that under the assumption of continuous rebalancing, the value of an option is independent from the expected return of the underlying asset but remains highly dependent on the expected volatility of the return of the underlying asset. The relationship between the percentage price movements of an underlying asset, and the percentage price movements of an option on the underlying assets is generally referred to as the "delta" of the option. In mathematical terms it is the first derivative of the option value with respect to the value of the underlying asset. In actual traded markets, assumption of continuous rebalancing is unrealistic. It would only make sense in theory if transaction costs were zero, while real world option trading is generally exposed to relatively high transaction costs. At the other end of the time horizon spectrum, the economic payoffs of options at expiration are well understood to typically have skew and other important features (even if the distribution of the returns of the underlying asset is normal). The question of how one would go about risk assessment and hedging of option positions over discrete time intervals has been less explored in the academic literature and is largely limited to cases involving the Black-Scholes option pricing model.
In this presentation we will provide a heuristic method that efficiently approximates the optimal delta hedge ratio for any finite time interval between now and the option expiration. The process is easily implemented for any option pricing model, including all options for whom the classic Black-Scholes model is not appropriate. It produces accurate option hedge ratios for the sort of finite time intervals (e.g. 10 trading days) that are often of practical consequence at the portfolio level, and that are consistent with many regulatory reporting regimes (e.g. UCITS). Most importantly, it allows for the risk of portfolios containing options to be assessed in a computationally very efficient fashion as compared to traditional Monte Carlo simulations as each option need be priced not more than four times. Finally, we will extend the process herein to address the question of the optimal rebalancing frequency of an option portfolio considering the related multi-period transaction costs as described in diBartolomeo (2012).
Private Equity at Center Stage: Maximizing Long Term Performance of a Diversified Portfolio Inclusive of Private Illiquid Assets
Presenters: Emilian Belev and Keith Black, PhD, CAIA CFA, CAIA Association
December 10, 2019
Abstract:
Any portfolio has two fundamental parameters that determine its suitability, regardless of whether it includes public or private investments. The first parameter is the projected performance over the expected investment horizon of the specific investor. The second one is the performance over horizons shorter than the full investment horizon, as the means to cover cash outflows from periodic liabilities and expenditures. The two objectives are related due to the fact that performance over long periods is a sequence of performance realizations over shorter term periods. In practice, however, the unobservable values of non-tradable assets pose a challenge to establish the relationship between their periodic and long-term performance. Our ultimate goal is to propose an approach for optimal total portfolio investing that maximizes long term performance, while also maximizing the investor confidence of covering periodic liquidity needs. Meanwhile we aim to address appropriately the complexity of a diversified asset portfolio without making assumptions about the unobservable aspects of performance of private assets.
To achieve these objectives, we first need to get a good understanding of all key features of the usual private asset investment vehicles used by investors. We have invited one of the most reputable experts on illiquid assets - Dr. Keith Black, who heads CAIA (Chartered Alternative Investment Analyst) Association research and curriculum - to start our presentation with a detailed overview of the private asset marketplace, the typical investment structures, and the current trends in this market.
Subsequently, we demonstrate how cash flow analysis can benefit the investor's quest for better transparency of the risk-return characteristics of private assets, while sidestepping the need for appraisal and accrual accounting derived returns. Our innovative cash flow analysis is based on collaboration between Northfield and Aspequity that combines forecasted expected investment cash flows with the application of Northfield's risk models, building a probability distribution of private fund cash flows - both inflows and capital calls. We show how the periodic probability distributions of cash flows aggregate into a cumulative cash flow statistical distribution over the lifetime of a private investment. The analysis demonstrates that the probability distribution of cumulative cash flows over the lifetime of private funds has non-normal properties. This puts under question some set-in beliefs about the valuation and deal flow management of illiquid assets, and sets the stage for new and superior approaches that are part of a currently forming standard in the industry. Finally, we outline a rigorous approach that utilizes the cash flow forecasting and risk models tools to deliver an optimized asset allocation consisting of both liquid and illiquid assets that maximizes long term expected performance, and meets an investor's liquidity targets over time with a chosen level of confidence.
Why a Total Portfolio Risk Model is not the Sum of the Specialist Models
Presenter: Emilian Belev
November 21, 2019
Abstract:
Forged between academia and industry, factor risk models evolved in one of two diametrical schools of thought. The first one developed with the advent of sell side factor systems which looked at individual instruments and focused on the variables that strictly define the value of the specific instrument type. For example, a model for a Spanish corporate bond of rating BB would strictly observe the "BB" yield curve for Spanish corporate bonds, a model for a call option on an index would use the implied volatility and spot price for that underlying, or a model for a swap would look at the swap curve for this particular instrument. This school of thought uses the combination of these variables as the genesis of the factor set across asset types. While the approach where specific pricing factors are reassigned directly as risk factors would be vastly superior than anything else at the individual instrument level, the quick compounding of the number of pairwise combination of factors would result in a significant number of estimation errors in the correlations. With a large number of factors where the common factors count increases roughly with the pace of new instruments in the portfolio, each new instrument introduces correlations with the pre-existing "common" factors and with that a corresponding number of estimation errors that get propagated throughout the portfolio.
On the other side of the spectrum are models with the least possible number of common factors. Admittedly they introduce larger errors when modeling individual instruments. However, when each such instrument is added to a mix of common factors plus idiosyncratic risk, the error introduced is predominantly on the side of asset specific variance, not the common factor variance. Correspondingly, such aberrations, being independent across instruments, will diversify away as we add more instruments. Hence the conclusion that in the presence of estimation error the number of common factors should be the possible minimum, which will minimize estimation error at the portfolio level. In the very extreme of this model design camp is a single-factor model like CAPM. The single factor model has been created as a theoretical assertion the assumptions of which do not credibly hold in the real world.
The balance between these two extremes is the paradigm which will serve the buy side asset managers of diverse asset class portfolios in the best possible way. While MPT does not give a formula to find the exact factors or number of factors, it is clear that the common dimensions of risk have to adhere to economic reason and should offer just enough explanatory power per single instrument to explain co-movement of value with other asset types, and no more than that. One technique to accomplish this task would be an explicit definition based on economic linkage and another would be a statistical technique like PCA. Northfield has chosen a hybrid method which combines the benefits of both.
The occasional concern that the risk bets of individual mandates will remain hidden to the enterprise risk function when the risk model is not a "sum of the parts" does not hold merit. Using factor mimicking portfolio techniques as an embedded feature of the enterprise risk system, the total portfolio risk manager can identify the breakdown of the specialist mandate risk bet and weigh against their realized payoff or the policy risk profile which the organization has as an objective.
On the Performance of Long-Term Momentum
Presenter: Jason MacQueen
October 23, 2019
Abstract:
Long-term Price Momentum, essentially based on the historic performance of a stock up to one month ago, has been a widely-accepted Style factor for many years, and is included in some form or other in most equity multi-factor risk models. It is also used by quants building multi-factor stock selection models and ETF managers running Style factor portfolios.
The usual belief is that high Momentum stocks outperform low Momentum stocks, or, as anyone old enough used to say, 'the trend is your friend'. However, although we have had a Long-Term Momentum factor in our XRD risk models for a number of years, it has become clear that the performance of the Momentum factor risk premium does not support this belief.
We have therefore been conducting some research into two different ways of defining Momentum betas, based on look-back periods of 6, 9 and 12 months, and on the performance of Momentum over investment horizons ranging of 1, 3, 6, 9, 12, 15 and 18 months out.
The results are somewhat surprising. We now understand why the Momentum factor in our risk models does not have a significant positive premium, and have also discovered a strong relationship between Size, as measured by market capitalisation, and Momentum. The only question we have not been able to answer is why fund managers are still using high Momentum to pick stocks.
Portfolio Construction Under Economic Scenarios
Presenter: Dan diBartolomeo
September 24, 2019
Abstract:
Many investment organizations spend a lot of effort to forecast future economic conditions. These forecasts are then utilized in various ways (often qualitative in nature) to influence decisions such as tactical asset allocation, and "macro-driven" changes in active portfolio strategies. The puzzling question for these organizations is how to accurately transform particular elements of economic scenarios such as forecasts of interest rates, exchange rates, trade levels, commodity prices, or consumer spending into explicit expectations of return and risk whether for asset classes, factor bets, active management styles (e.g. "value", "momentum), or individual securities.
In this webinar, we will present how our method of "optimized scenario analysis" in conjunction with our Everything, Everywhere model can be used to translate economic forecasts directly into forecasts of return, volatility, skew and kurtosis for any individual security or any set of securities making up an asset class (i.e. a index portfolio). Complex economic scenarios with many different elements are supported across countries, regions, or globally. Any number of entire scenarios (e.g. "recession", "expansion") can be combined over a user specified time horizon for economic events.
A key benefit of the process is that the variables being forecast can be any economic or financial market measure with available historical time series data. The scenarios are not limited to items that are specified as factors in any of our models. The four moment descriptions of the asset return distributions can then be used as inputs to our Open Optimizer which has recently been enhanced to incorporate the influence of higher moments in determining portfolio allocations while retaining all existing Optimizer functionality.
Methods for Joint Optimization of Multiple Related Portfolios: Householding and Beyond
Presenter: Dan diBartolomeo
August 29, 2019
Abstract:
Then many cases, both investors and asset managers would benefit from the ability to jointly optimize over an entire set of related portfolios. This presentation will focus on the theory and implementation of three applications of joint optimization.
In the wealth management area, the multiple portfolios by various members of a specific household (parents, children, retirement accounts, education savings, trust funds) can be separately optimized to reflect the differences in financial goals, risk aversion and tax circumstances. However, any synergies across portfolios that might be possible with a more holistic process are lost (see Fowler and deVassal, Journal of Wealth Management, 2006). Alternatively, the wealth from these many accounts could be pooled into a family fund to be optimized as a single larger portfolio. However, the rational parameters of the pooled problem will be dominated by the higher value portfolios within the set, while being less influenced by the smaller value portfolios. As such, the pooled solution may be "unfair" to some members of the household even if it is Pareto optimal for the family as a whole. Northfield recently introduced a very robust "householding" capability into our line of portfolio optimization applications, which allows for any desired degree of compromise between what is the set of separately optimal portfolios and the optimal pooled portfolio.
A second important application of joint optimization capability is the formation of investment portfolios of complex corporate entities, such as multi-line insurance companies. Such companies may have dozens or even hundreds of separate portfolios designated as the investments for a particular line of insurance in a particular jurisdiction (e.g. car insurance in Texas). Each of the separate portfolios may have its own tax, financial and regulatory considerations. Multiple portfolio joint optimization allows for the many portfolios to be optimized in a holistic fashion maximizing tax and other synergies across the entire enterprise, while conforming to constraints associated with the separate sub-portfolios.
A third use case is that of a traditional asset manager who may manage "many similar but not identical" portfolios for related parties. For example, many large state pension funds also manage money on behalf of smaller units of government such as towns or counties.
Multiple portfolio joint optimization can be usefully applied to many common problems such as how to allocate an equity share purchase across investors when the optimal total number of shares cannot be obtained due to liquidity limitations.
On the Value of Portfolio Construction
Presenter: Jason Macqueen
June 25, 2019
Abstract:
Active portfolio management consists of two steps: stock selection and portfolio construction. Many managers spend most of their time on stock selection, and then deal with portfolio construction by following some simple heuristic, such as equal-weighting or capitalization-weighting. Most ETFs, for example, use one of these methods.
Unfortunately, this is virtually guaranteed to result in inefficient portfolios that do not trade-off expected return against risk, and which may well have exposures to significant unintended bets. The managers stock selection skill can easily be dominated by the returns to these unintended bets.
In this presentation, we use a simple stock selection rule (similar to those used to create some Style ETFs) and test a number of different methods of portfolio construction. These will include equal-weighting, capitalization-weighting, attribute-weighting, risk parity weighting and Markowitz optimization. Each of the strategies is rebalanced quarterly from the end of 2005 to the present. Since the stock selection is always the same, the differences in performance and turnover are due entirely to the different methods of portfolio construction. You may be surprised by the results!
Bad Benchmarks, Passive Investing and ETFs
Presenter: Dan diBartolomeo
June 25, 2019
Abstract:
The last decade has brought about a massive increase in the use of passive strategies by both institutional (index funds) and retail investors (ETFs). In turn, this has lead to a material proliferation of thousands of "benchmark indices" that describe some way of purportedly investing passively in some defined set of securities. While many of the concepts and strategies embodied in these benchmark indices are quite sound, many indices now being marketed by index data providers have material conceptual problems. Many indices are very poorly suited to actually fulfill the purposes to which indices are routinely put. Among these purposes are to act as a representation of an asset class for the formulation of capital market assumptions, the ability to act as a fair benchmark for active managers, and to be the basis of actual passive investments such as index funds and ETFs.
In this presentation, we will show that many indices cannot possibly fulfill all these requirements and what can be done to mitigate the various problems that arise.
The massive number of indices has created the bizarre situation that multiple indices routinely exist representing the same opportunity set (e.g. the S&P 500 and the Russell 1000). How can it be that an index fund would be judged a failure if the fund missed tracking the underlying index by more than a few basis points, while two indices purportedly representing the same asset class have tracking errors of 100 basis points or more to one another? In purportedly liquid equity markets, government ownership of enterprises (e.g. China) or in some cases entire economic sectors (e.g. oil in many Middle Eastern countries) are not addressed, and there are assumptions of non-existent liquidity (i.e. there are lots of ETFs where the trading is orders of magnitude higher than the liquidity of the underlying securities). In the fixed income world, there are a number of problems including application of equity-style market value weighting leading to the perverse condition that the deeper a borrower gets in debt, the more you should lend them. Among illiquid assets there an equal number of logical inconsistencies leading investors to make bad investment decisions. For example, the cross-section of returns in private equity and venture capital have extreme positive skew, meaning that the average return to any asset class index will be far higher than the median of the distribution. As such, investors must be "hyper diversified" to an impractical level just to achieve the average return. Illiquid indices also exhibit "appraisal bias," which overstates returns and dramatically understates volatility. Finally, real estate indices cannot actually be held by investors as most real estate properties are indivisible. Each parcel of real estate included in the index is entirely owned by a specific investor, which means that same parcel cannot be owned by other investors.
Illiquidity Risk of Truly Illiquid Assets
Presenter: Emilian Belev
May 30, 2019
Abstract:
It is a very common perception that investors dislike being constrained by liquidity of their investments at their investment horizons. There has been a notable amount of academic and practitioner research work to try and capture the disutility of illiquidity and its implication to risk and return. Most of those studies have been done form the perspective of endogenous variables that measure this disutility - expected return premium for illiquidity or inferred from bid-ask spreads. Several years ago we presented an approach based on a simulation of liquidity constrained optimizations as another endogenous approach to liquidity risk estimation.
In this work we present an exogenous, transaction frequency based approach, that is economically synonymous with our prior work, but empirically more tractable, granular, and transparent. We will use examples from private equity and commercial real estate to illustrate the new methodology.
One of the implications of this work is an explicit way to measure the expected cost of risk for asset types that don't have well defined data for bid-ask spreads.
Did the Global Financial Crisis Change Equity Markets for the Better or Worse?
Presenter: Dan diBartolomeo
April 30, 2019
Abstract:
The Global Financial Crisis of 2007-2009 is widely considered the most important event in financial markets since the Great Depression of 1929. Now that roughly a decade has past since the this catastrophic even, investors and asset managers must ask themselves what has changed in the aftermath of the GFC, and what new opportunities and risks have evidenced themselves?
To answer these questions we will examine the extensive data from Northfield's Corporate Sustainability Model as described in diBartolomeo (Journal of Investing, 2010). In this variation on the contingent claims analysis approach (Merton, 1974), the rights of a corporate shareholder are described as a call option and a put option on the assets of firm. The key input to the process is the volatility of the firm's assets, which is equivalent to "How volatile would the firm's equity be if the firm had no debt?" By equating the value of the two option portfolio to the firm's stock price, we can solve for the implied expiration date of the options which is the market expectation of the survival time of the firm. The Northfield dataset derived from this measure starts in 1992 and continues to the present day for all US companies and non-US companies traded in the USA in ADR form.
In the presentation we will compare the data from before and after the GFC both in market wide summary and at the granularity of sectors so as to illustrate what has and has not changed in terms of corporate sustainability, and in particular the stability of the financial system. The differences from before and after the GFC indicate material change in some but not all sectors of the economy. These changes imply differences in sector level equity returns, and variation in the expected returns to equity investing styles and factors such as "value" and "quality."
Introducing Northfield's Global Pure Fixed Income Model
Presenter: Emilian Belev
March 26, 2019
Abstract:
In this presentation, we will discuss the concepts, structure, and innovation behind the new Northfield pure Fixed Income Model.
The discussion will include comparison with industry's established practices in this area, and how we adopt or improve on them.
Specific details will be shown in terms of model performance, coverage, and benefits.
Dealing with Japanese, Frontier Equity Markets by Considering Non-Traditional Distributions
Presenter: Dan diBartolomeo
February 26, 2019
Abstract:
Global equity markets had materially negative returns during the fourth quarter of 2018. Among the largest magnitude declines was the Japanese market where the index values lost almost 20% during this period. Institutions that held large positions in Japanese equities have publicly reported massive losses with one institution reporting a Q4 loss in excess of US $100 Billion.
In this presentation we will illustrate how proper consideration of non-traditional return behavior (skew, kurtosis, serial correlation, correlation jumps) should impact our understanding of the risk of various national equity markets.
It is customary in the investment industry to make certain assumptions which make asset allocation based on Modern Portfolio Theory (Markowitz, 1952, 1959) tractable. Typical assumptions include that returns are normally distributed, volatility is constant and returns are not serially correlated. While these assumptions are clearly untrue in high frequency data (see diBartolomeo, Professional Investor, 2007), most empirical research would suggest that "normal and IID" assumptions would not be routinely rejected for returns when measured over longer periods such as months or calendar quarters. However, there are many financial markets where analysis of historical return data even at monthly periodicity would reject the usual assumptions. One clear example of non-conforming behavior would be the Japanese equity market which has been subject to powerful long-term trends for nearly four decades. Similarly, many emerging and frontier markets have demonstrated very extreme return events (e.g. Zimbabwe, Venezuela) far more frequently than would be expected.
The key outcome is that the expected frequency of extreme events impacting global equity investors is far higher than is typically estimated.
Accounting for Heterogeneity of Security Behavior with "Bottom Up" Asset Allocation
Presenter: Dan diBartolomeo
January 24, 2019
Abstract:
In this presentation we will show how estimating asset class volatility and correlations from the "bottom up" (i.e. from current security compositions) provides material advantages relative to traditional time series methods when applied to collective assets.
The dominant basic construct for allocating to the various assets within a portfolio is the classic Markowitz Modern Portfolio Theory (1952). This process requires formulating expected returns and covariances among the assets defined. An important nuance that most investors overlook is the distinction between allocation among singular assets (i.e. individual securities, currencies, etc.) and collective assets such as the set of singular assets included in a financial market index, which is then used as a representation of an asset class.
While many techniques have been developed for forecasting the parameters of the allocation problem, most are applied uniformly to both singular and collective assets. This omission materially increases estimation error for common asset (and ETF) allocation problems where collective assets are the primary vehicles. For example, consider estimating the ex-ante volatility of the S&P 500 as of December 31, 2018 (after a sharp drop of around 11% during December). Time series approaches such as ARCH and GARCH would suggest that ex-ante volatility estimates should increase in the wake of a volatile period, but the heterogeneity of the securities within the index plays a material role. In a sharp decline, high volatility (high beta) securities fall more and low volatility (low beta) securities fall less in value. Accordingly, the riskier securities are now a smaller portion and low risk securities are now a larger portion of this capitalization weighted index. This shift of weights within the index composition will mute the expected increase in volatility. A sharp increase in market valuation would have the opposite effect, increasing the weight of riskier securities thereby expanding any expectation of increased volatility. In relatively concentrated indices (smaller markets or high average correlation) this effect can be sufficiently strong to actually decrease expected volatility after a large market decline, contrary to most investors intuition. The same mechanisms impact the expected correlations across asset classes represented by market indices (see our September 2002 newsletter).
The same heterogeneity effects can also be important in asset allocation of taxable portfolios as described diBartolomeo (2003, 2008) and Markowitz and Blay (2016).
Dissecting Duration Times Spread
Presenter: Dan diBartolomeo
December 27, 2018
Abstract:
In recent years, the “Duration Times Spread” (DTS) methodology has become the most commonly used approach to estimate the risk of bond portfolios. DTS was first formalized by Ben Dor, et. al. (Journal of Portfolio Management, 2007). Put simply, it suggests that the volatility of the portion of bond yields associated with credit and liquidity risks (i.e. OAS) is linearly related to the magnitude of the spread itself. Effectively what matters is the percentage change in the spread (e.g. going from a 200 bps spread to a 400 bps spread is 100% change). This functional form implies that volatility of spreads will be linearly related to the level of the spread, so when spreads increase so will the expected volatility of the spreads and vice versa. While this construct fits empirical data very well, there are a number of limitations to this approach of which investors must be aware.
The first matter of interest is that the arithmetic value of the “duration times spread” is exactly the maximum amount by which a bond portfolio can increase in value due to the portfolio becoming riskless with respect to creditworthiness and liquidity. Most investors would find the idea of measuring risk by how much a portfolio can increase in value unintuitive. Implicit in the idea that a measure of “upside” should be considered an estimate of risk is that the distribution of credit related returns for a bond portfolio is assumed to be symmetric. We would clearly not expect the returns of a single bond issue subject to default risk to be symmetric. Under the Central Limit Theorem, we can justify not only the expectation of symmetry but also normality for the distribution of returns. However, the requirements of the CLT are that we are summing a large number of independent return distributions (i.e. the returns of individual bonds). To the extent that the creditworthiness and liquidity of the bonds in a particular portfolio are likely to be correlated rather than independent there must be a degree of portfolio concentration such that the CLT requirements are not met and the expectation of symmetry must fail. This situation is similar to the assumptions made under the Gaussian Copula process for debt securitizations by Li (Journal of Fixed Income, 2000) which were subject to material criticism in the “Global Financial Crisis” period.
The second limitation of DTS is that as the creditworthiness of a bond falls investors must eventually reach a point where there is a high degree of certainty that a particular bond will default. In such a case, the value of the spread should increase (the expected loss is increasing due to the likely default), but the volatility of the spread should decrease as investors reach consensus that the bond will indeed default. This situation contradicts the basic DTS premise that the volatility of spreads should be linearly (positively) related to spread levels.
Finally, we would note that all empirical research on fixed income securities may be limited in the sense that the widely disseminated prices for bonds are derived estimates typically from “matrix pricing” models with limited amounts of actual trade data as input. As such, the in-sample explanatory power of any such model may be materially overstated because the matrix pricing model and the risk model simply have common underlying assumptions.
This webinar will consider all these limitations and particularly discuss how they relate to traditional credit risk concepts like probability of default (PD) and loss given default (LGD).
Why Active Managers Should Not Try to Maximize IR or Use Tracking Error as a Risk Measure
Presenter: Dan diBartolomeo
November 27, 2018
Abstract:
Active managers must believe they will outperform their benchmarks in order to rationally take up the task of being an active manager. However, it must be arithmetically true that for every investor who outperforms fair benchmarks there must be other investors who expected to outperform, but actually underperformed. It cannot be true that all managers will be above average. As such, we have the paradox that the intrinsic positive alpha expectations of roughly half of all active managers must be wrong. The conventional view of active risk is a tracking error estimate around an expectation of a positive alpha which does not reflect this clearly embedded “wrongness.” In essence, we have the perverse statistical description of a dispersion (standard deviation) around an unknown location (future mean alpha).
To resolve this issue, we propose a new measure of active risk where the distribution of active returns is bi-modal with two modes with non-zero probability at both positive alpha and negative alpha. Relative to the investor expectation of positive alpha, this distribution has negative skew and positive kurtosis.
The computational process is formulated in two steps. First, we treat the distribution as a mixture of two normal distributions, given the probability P of being centered at positive alpha, and (1-P) being centered at negative alpha. We then apply, the Cornish Fisher (1937) expansion method to convert the four moments of this distribution to the tractable equivalent of a single volatility metric which is the proper assessment of active risk. An important theoretical result that arises from this construct is that maximizing the information ratio is a poor objective for active managers, due to the highly non-linear relationship between tracking error and the more robust active risk estimates.
Face Off: The Factor Model vs. the Commercial Real Estate Risk Premiums
Presenter: Rick Gold and Emilian Belev
October 30, 2018
Abstract:
It has been more than a decade since the first generation of Northfield’s real estate factor model was created. It was founded on the premise that granular and specific inputs near and dear to real estate investors can be leveraged to facilitate the creation of factor models that are also intuitive to risk managers. Since the model’s debut, we have used our experience to extend the model’s coverage, add enhancements, and expand its application.
One of the challenges of factor-based private assets modeling is to demonstrate that the approach works given limited transparent arms-length pricing and return data and most important, the well documented limitations of traditional real estate appraisal-based benchmarks. We have accomplished this in large part by devising methodologies that allow us to make reasonable comparisons against well-known non-arms-length pricing data (appraisal-based indices).
In this presentation, we make use of real estate transaction-based capitalization rate data to draw inferences about the performance of our existing model, and by extension, the potential of such data to add value to the model’s estimates. While we have found that the model does a great job at on its own, the cap rate data does provide additional improvements in fine-tuning cap rate spreads compared to the model’s discount rate. This is welcome news since by its nature the incorporation of cap rates fits seamlessly in the model’s existing structure. It does so in a way that respects the granularity of the model and respects building and location specifics, rather than taking a “one-size-fits.”
The Fundamental Myths of Fundamental Models
Presenter: Dan diBartolomeo
September 27, 2018
Abstract:
Many investment organizations have come to the belief that "fundamental" (endogenous) models of equity returns and risk are somehow more useful than other models for a variety of reasons. While our US Fundamental model and XRD line of international equity risk models are extremely effective, we have noted that some investment organizations have only a vague understanding of the true advantages and disadvantages of such models.
For purposes of definition we will consider a "fundamental" model as one where the factor exposures at the security level are observable at a moment in time. The analytical structure of the model then involves statistical estimation of factor returns for each time period, and the estimation of the covariance matrix of these factor returns. By way of example, we might think of the capitalization of a company, or the price/earnings ratio of the stock, or the membership of the firm in an industry group as "fundamental characteristics" which we can observe at the security level. There is no need for statistical estimation of the information describing individual stocks or companies. Often, the "fundamental" aspect of such models arising from the use of data elements from the firm's financial statements (i.e. book/price ratio) which is in some measure parallel to the way that fundamental investors might view a particular equity security. We find that there are important aspects of fundamental models which are widely misunderstood. In particular, the assertion that fundamental models are inherently more accurate than other risk model structures is entirely false.
The purpose of this presentation is to clarify the nuances as to what is and is not real about fundamental models, as compared to model frameworks of return and risk. Assessing the benefit of endogenous models requires not only a full understanding of the statistical properties of such models, but also context of usage in terms of the investor's strategy and the relative importance of forecasting benchmark relative or absolute risk.
Fulcrums, Levers, Pulleys and RAMP: The Northfield Analytics Framework and SAAS Platform - Modern Tools for Risk Analytics
Presenter: Daniel Mostovoy
September 11, 2018
Abstract:
This presentation will illustrate how financial institutions such as asset owners can enhance their risk management using an online system that is very convenient (e.g. hosted by Northfield) while offering an unprecedented level of flexibility and data security. In 2015, Northfield introduced the planned concept of RAMP, an entirely new way of implementing risk reporting and optimizations. In recent years the issue of widespread "data hacks" have had broad impact across financial institutions and society in general.
What makes RAMP unique is that the portfolio holdings of financial institutions are never stored at Northfield, removing the potential for data hacks causing disclosure of confidential investor information.
RAMP queries the original source of the data (e.g. custody accounting systems), loads them into data structures in memory on a secure server, produces the required analytics, then immediately deletes all the input information, without ever having stored it. Output reports are distributed through secure processes. This is accomplished using a software framework where analytics processes can be prototyped quickly and put in production as a service.
At full implementation, the RAMP environment will offer clients a wide range of Northfield analytics including risk reporting, scenario analysis, stress testing, optimization and our enhanced report suite (now available in MARS- ERM).
Parameterization of the Tax/Risk Tradeoff for High Net Worth Investors
Presenter: Dan diBartolomeo
August 28, 2018
Abstract:
In the more than twenty years that Northfield has offered tax sensitive optimization techniques, many of our clients have struggled to get high net-worth clients to pursue economically rational tradeoffs between investment related taxes and portfolio risk. While reductions in taxes can be easily demonstrated when investors make smaller tax payments than they otherwise would have done, the benefit to investors from reducing portfolio risk accrue slowly over time and perceived to be beneficial only under volatile market conditions.
In this presentation, we will provide the explicit framework that provides the most transparent method for sensibly making decisions for tax/risk tradeoffs. While the basic structure of the problem has been in place since the introduction of our tax optimization methods in 1996, we will explore many nuances to the appropriate parameterization of the problem. Among these nuances are the explicitly multi-period nature of the problem, correctly defining portfolio turnover for active management, and the inclusion of investor-centric issues such as mortality risk and charitable giving.
The Second-Order Risk of Portfolio Factor Bets
Presenter: Dan diBartolomeo
July 24, 2018
Abstract:
It is widely asserted that active managers should avoid "inadvertent factor bets," but few people recognize all aspects of this concept. What is often missed in the analysis is the potential performance impact of times series variation in accidental factor bets. There are a couple strands of logic behind the idea. The first is that any active factor bet is a source of risk in terms of benchmark relative performance. Clearly, if the factor bet is accidental we should have no expectation that it will enhance performance, so we are taking risk for no economic reward. However, the concept of an active factor bet is typically considered only at a moment in time. The second rationale is that "you only get hit by a car you don't see coming," so the potential for harm from an inadvertent bet is conceptually greater than from an intentional active bet. Even if we are factor or industry neutral on average this does not guarantee that we are neutral at each moment in time.
Consider the example of a portfolio that has no particular "bullish" or "bearish" outlook on the market and so intends to have a market beta of one to the benchmark. However, if by virtue of security selection decisions the portfolio beta varies randomly over time from .8 to 1.2 there will be a very meaningful impact on active performance even if the portfolio beta is exactly one on average. In this presentation, we will illustrate the performance impact of random variation in the magnitude of factor bets. We will show that in many cases the random time series variation in factor bets represents a far greater influence on realized performance than having a fixed level of those same factor bets, even if inadvertent. To the extent that more frequently rebalancing of portfolios can reduce both average factor bets and also the random variation in bet magnitudes over time the potential improvement in return/risk tradeoffs must be carefully weighed against the performance drag of greater transaction costs.
How to Build Smarter Portfolios Using Factor Models
Presenter: Dan diBartolomeo
July 10, 2018
Abstract:
As institutional portfolios grow in size and complexity, factor models play increasingly important and versatile roles across the buy-side investment process. Join our panelists, Dan diBartolomeo, President, Northfield Information Systems and Katya Taycher, Director of Product Management, Charles River, for a discussion moderated by Frank Smietana, Product Marketing Manager, Charles River. Topics include how to:
· Embed models to gain timelier insight into risks and opportunities
Generating Tax Alpha
Presenter: Steve Dyer
June 26, 2018
Abstract:
After a decade of consistent upward market movement in the US, private wealth managers are finding themselves in the position of handing client's large tax bills or trapped with a locked-up portfolio full of low cost basis holdings. It's great to be able to tell your clients their portfolios have performed well, but managers are also asking how these same accounts would have performed had they been more carefully managed to lower tax bills.
Tax alpha consists of two parts: Using capital losses to offset gains and reduce the overall tax bill, and deferring paying taxes as long as possible to allow the returns to the portfolio to compound before realizing gains.
The ability to offset gains with losses is mainly predicted by the overall market returns and the cross-sectional dispersion of the universe - even when the market is going up, there are individual stocks that are going down and creating opportunity to tax loss harvest. The value of deferring paying taxes will vary with the age of the portfolio and the client's comfort with allowing the portfolio to drift away from the agreed upon strategy.
In this webinar, we will present the results of an optimized backtest on a passive US equity strategy to quantify the amount of tax alpha realized under different time horizons, market conditions, and universes to demonstrate the value of incorporating tax-sensitive strategies into your management practice.
Maximizing RAROC: Turning the “Risk” Unit into a Profit Center
Presenter: Emilian Belev
May 29, 2018
Abstract:
The asset management industry has long been endowed with optimization tools that source their roots in Modern Portfolio Theory which lets risk factors drive returns in a fine-tuned fashion. Banks and other financial institutions whose direct revenue driver is not necessarily AUM, on the other hand, have historically considered risk mostly from the perspective of the binary dimensions "solvent" (or not) and "regulation compliant" (or not). Due to this assigned role, risk has been able to demonstrate its value only under doomsday scenarios or after successful rounds of regulatory review, but not explicitly in the bottom line of a non-eventful period's profit report. This has allowed politically misinformed language to build up in that space which equates the risk function to increased business cost and impediment to growth, and not to an opportunity.
This presentation dispels this erroneous notion that is harmful to an organization's profitability. It further unveils a new framework that is based on award winning work that offers a clear, rigorous, and effective method to optimization of risk adjusted returns that will give risk managers the tools to be the stars of their own show. he approach applies equally well to an asset manager's portfolio, as well as to a bank's trading or banking book, incorporating distributional moments of all orders while adhering to a preference function that is close to heart to both bankers and money managers. As being distinct from the commonly assumed exponential utility form, it offers a viable alternative that can replace, where the latter is not applicable, or complement, conventional MVO optimization results. Capturing explicitly the effect of leverage, it is also an attractive framework for hedge fund managers.
Introducing MARS-ERM
Presenter: Emilian Belev
May 17, 2018
Abstract:
MARS-ERM is Northfield's complete risk management infrastructure for institutional asset owners. It provides a 360-degree view of your entire risk profile with modeling for all asset types globally, including illiquids, derivatives, and structured fixed income and uses a thorough, and innovative approach.
Emilian will introduce the audience to this standalone enterprise risk solution, its objectives, operation, and reports, as well as a quick overview of the underlying model analytics.
Return and Risk in Endogenous Time
Presenter: Dan diBartolomeo
April 26, 2018
Abstract:
This presentation will explore one of the most basic yet least understood concepts in all of investing, the nature of investment return.
We define the return on an investment as the rate of change in the value of the investment per unit time, which ought be understood as a simple arithmetic ratio. To a nearly complete extent, the attention of both investment practitioners and finance academics have been focused on the numerator of the ratio, the change in the investment value (price). It has always been the custom of the industry to take the concept of units of time as a given and immutable quantity.
We assert there is material advantage conferred by analyzing the usually neglected denominator of the ratio. For example, it is quite common for equity traders to think of time as being measured more in trading volume, rather than the standard units of seconds, minutes and hours. This reflects knowledge that some times of the trading day are routinely very busy (around the open and close) while other parts of the day are typically slow, with reduced trading activity. In terms of asset valuation, hedonic models are very common in which an asset may be described as "overvalued" or "undervalued" without regard to how quickly the market price and estimated value are likely to converge. Within the realm of technical analysis there is the "point and figure chart" which entirely eliminates the concept of time in describing price change patterns.
Our Risk Systems That Read® method for improving risk estimation treats the passage of time as varying according to the amount and importance of information coming to investors. The idea of time itself speeding up and slowing down has been previously explored in papers such as Derman (2002), Haug (2004), and Kyle, Obizhaeva, Sinha and Tuzun (2012). One important strand in this literature is the process by which speculative "bubbles" in prices may arise.
The Northfield XRD Risk Models - A Different Perspective
Presenter: Jason MacQueen
April 10, 2018
Abstract:
The XRD risk models are built somewhat differently from other Northfield equity risk models, and this webinar will review the important differences.
Whereas most medium-to-long horizon models are based on a single sample of calendar month returns, the long-term XRD model uses four sets of 4-weekly returns, while the short-term XRD model uses five sets of weekly returns.
The factor structure of the XRD models is more granular than other models, enabling fund managers to look at the risk from 20 industry groups, for example, rather than just a small number of broad sectors.
The XRD models also include a number of Style factors, derived from stock fundamental data, and we will be discussing the evidence they provide of the persistence of various Style factor premia.
Finally, we will also discuss different ways in which a Size effect might be reflected in a risk model, and discuss the method used by the XRD models.
The Liquidity Risk Time Bomb
Presenter: Dan diBartolomeo
March 27, 2018
Abstract:
During the financial crisis years of 2007-2009, much of the declines and volatility experienced by global markets purportedly had to do with liquidity related concepts. These effects ranged from the "hedge fund meltdown" of August 2007 to the destabilization of money markets triggered by the failure of Lehman Brothers. The near-failure of numerous other financial institutions contributed further to the misery, to which central banks responded with unprecedented injections of massive liquidity into financial markets.
Since then, the equities world has been subjected to lots of discussions on "crowding" of strategies and factors. Interest rates have gone to zero or even negative in many countries. The rapid growth of ETFs makes the current problem worse, as ETFs are traded with high liquidity but without regard to the fact that many of the underlying securities may not be equally liquid. Regulators such as the US SEC and the various aspects of MIFID II in Europe have begun to require that asset managers of open-end funds and ETFs carry out analyses of their liquidity risk. In addition, regulators desire trading practices that do not unfairly shift the cost burden of large liquidations to remaining investors from those investors withdrawing.
In this presentation, we will describe various approaches that different participants in the asset management industry are taking to analyze liquidity risk. Unfortunately, we find that in all but a few cases the analytical approaches being undertaken are unsound. These flawed analyses give the impression that market liquidity to transact securities is far greater than it actually is.
We will conclude with an optimization based approach for managing liquidation costs in crisis conditions.
Why Getting Risk Right is Wrong
Presenter: Dan diBartolomeo
February 27, 2018
Abstract:
Many investment professionals who use risk models make a common mistake. They assume that a risk model is working well if the amount of volatility realized by a particular asset or portfolio is consistent with what the model had predicted. They believe that volatility forecasts should be an unbiased estimator of subsequent realized volatility.
In this presentation we will provide five different rationales as to why seemingly unbiased estimates of volatility are undesirable both statistically and economically.
The implications of these arguments are that professional investors routinely take too much risk, back-tests and simulations fail to capture the true risk of strategies, and that evaluation of investment performance is biased toward perceiving luck as skill -- leading to upward biased performance related compensation.
Optimization 101
Presenter: Steve Dyer
February 15, 2018
Abstract:
Optimization is just the process of making something the best it can possibly be. At Northfield, we hope you are using our optimizer to make the best investment portfolios you can, but optimization has hundreds of applications across almost all industries.
This workshop will be an introduction to the concept of optimization and a discussion about Northfield's portfolio optimization methodology, and how you can use the tool to consistently and efficiently implement your investment skills and best thinking to your portfolio, or to dozens of strategies across thousands of accounts.
This workshop assumes no prior familiarity with Northfield or optimization, and is meant for those new to Northfield, as well as to people who have roles such as marketing, development, or IT in organizations that use Northfield who would benefit from having a working knowledge of the service.
It will be accessible to all professionals with no math beyond high school algebra.
Investment Management for Private Equity and Venture Capital General and Limited Partners: Risk, Return, and Diversification
Presenter: Emilian Belev
January 23, 2018
Abstract:
With its appeal of early insight-driven performance, combined with yield-hungry investors in a low interest rate environment, private equity has attracted vast amounts of capital and interest. Due to the inherent focus on keeping investment insights private to preserve their value for the benefit of both GPs and LPs, the requisite data has not been readily transparent to subject the asset class to the same quantitative analytical approaches as its public equity counterpart, leaving a great deal of "take it or leave it" risk for investors.
In this presentation, we assert that this should not be a defining limitation of the asset class. Armed with only innocuous information about the portfolio investments and the manager's past performance, a well-defined expectation of future performance and risk can be established for the partnership. This analysis can then be seamlessly integrated in a methodical approach to optimal illiquid asset investing.
In addition to the obvious benefits to limited partners, general partners will have benefits of their own. First, they will have a way of showcasing their superior outlook in a competitive environment, while not spilling the "secret sauce." Secondly, they will also have a way of keeping track of the contribution of performance on a risk-adjusted basis by individual general partners and mangers, which will help them attract, reward, and keep talent central to their sustained performance.
Introducing Bootstrap Scenario Analysis in Northfield's PRISM System Portfolios
Presenter: Richard Dawson
January 9, 2018
Abstract:
This workshop will offer a brief introduction to Northfield's PRISM system.
PRISM is an intuitive and powerful client-server based application, providing automated data management and risk reporting, leaving the analyst free to concentrate on the problems that concern them. It is compatible with all Northfield risk models and data.
We'll be focusing on the Scenario Analysis package in this talk. Up to now this has only been available in our standalone NISBoot application. Now available through PRISM, this is a unique approach to optimized, scenario driven risk analysis. Developed to resolve the shortcomings of numerical methods we have built a new process, extending the approach suggested in Meucci (2008) which combines Monte Carlos simulations with the flexibility to overlay complex explicit scenarios.
The computational process involves an optimization problem that calibrates our "bootstrap" resampling process (see our newsletter June 2013) to one or more user defined scenarios. The analytical output of the process is a robust representation of the distribution of possible outcomes, while being consistent with any mathematically feasible "stress scenario."
Smart Beta Corporate Bond Portfolios
Presenter: Jason MacQueen
December 28, 2017
Abstract:
In October 2015 we held a webinar on Smart Equity Portfolios. Although Smart Beta ETFs have become very popular, our contention was that the way in which the portfolios were constructed was not very efficient, and that the actual performance of such funds were therefore driven as much by their exposures to other factors as they were by their exposure to the target Style factor.
We created a number of optimized Smart Portfolios, in which we deliberately maximized the exposure of each portfolio to the target Style factor, while minimizing its exposure to all other factors as far as possible, consistent with the long-only constraint. These Smart Portfolios' performance compared very favorably with many of the Smart Beta ETFs available in the market at the time.
In this research exercise, we are looking at using the composition of the Smart Equity Portfolios to build a set of corresponding Smart Corporate Bond portfolios. The holding of each equity is replaced with a corporate bond issued by the same company. To do this, we use the Merton formulation of a corporate bond as effectively consisting of a combination of the underlying equity and some (risk-free) treasury bonds.
Complete Attribution for Quantitative Managers
Presenter: Leigh Sneddon
December 12, 2017
Abstract:
The choice of signals, their weights, and the portfolio constraints, are decisions at the core of the portfolio manager's role. Both managers and their clients need to know how successful those decisions have been. Traditional attribution systems, however, commonly leave half of portfolio performance unexplained and deliver counterintuitive "explained" results, and many do not attribute to signals or constraints at all. This leaves portfolio managers and clients caught between a rock and hard place.
Complete Attribution (CAtt) solves these problems. It attributes risk and return to core investment decisions, particularly signals and constraints. It attributes all performance: there is no re-distribution or unexplained plug ("stock-specific" etc.). CAtt differs from traditional systems: it is not Brinson-like, does not attribute to risk models, and does not use regression.
CAtt works by closely tracking the investment process. It uses all the inputs to portfolio rebalancing, and models the portfolio rebalancing process in detail. Proprietary data stays at the manager's location. Importantly, CAtt captures portfolio dynamics: the impact of each day's signals and constraints cascading into the future. It is cause-and-effect attribution, tracking the impact of each day's signals and constraints on current and future risk and return.
The presentation will illustrate how CAtt answers important questions on the impact of individual signals, the risk budget, contributions of new signals, turnover selection, trading costs, signal performance trends, event analysis, constraints, Smart Beta and factor timing.
By delivering reliable, residual-free attribution, CAtt provides competitive advantages to asset managers. The insights both promote better future returns and raise the level of discussions with prospects and clients. For the first time, active quants can attribute to their investment decisions and get results they can believe and use.
The Lack of Market Volatility
Presenter: Dan diBartolomeo
November 30, 2017
Abstract:
Since the end of the Global Financial Crisis generally, and in recent months even more so, financial markets around the world have experienced unusually low levels of volatility.
In this
presentation we will describe several different conditions which we believe
contribute to this situation. Most importantly we will present a new theory
under which the current high levels of international and political tension
around the world reduce current market volatility rather than contribute to it.
The presentation will focus on the differences in how investors respond to news
(an unanticipated event), scheduled announcements (e.g. an earnings release)
and events which are considered inevitable but unknown with respect to time
(e.g. the eventual eruption of an active volcano).
The final portion of the presentation will discuss the implications for
tactical asset allocation and active portfolio strategies.
Risk Models 101
Presenter: Steve Dyer
November 16, 2017
Abstract:
Are you completely new to using Northfield? Do you think of risk in a framework that isn't through a lens of multifactor risk models, or otherwise outside of what is considered "quant?" Are you in a role that is on the periphery of Northfield services, such as marketing, development, or IT, that would benefit from a working knowledge of Northfield services?
Then please join me for this 30 minute introduction to risk models. It assumes no prior background and is meant to be accessible to all professionals.
We will answer simple questions to get you started:
Really Sustainable Long-Term Investing
Presenter: Dan diBartolomeo
October 24, 2017
Abstract:
In this presentation we will examine the attributes that allow companies to survive in the long run. Through the lens of the nearly thirty-year history of the Northfield risk models we can examine the set of companies, both in the US and globally, that have survived a quarter century or more.
In the first part of the analysis, we contrast the characteristics of the portfolio of survivors with the portfolio of non-survivors at various time points over the past three decades.
In the second part of the analysis, we compare the set of actual survivors with the set of expected survivors as predicted by the sustainability model of diBartolomeo (Journal of Investing, 2010).
The final part of the analysis will be to review the market-relative performance of the portfolios of both the predicted and realized sets of "sustainable" firms so as to consider the investment viability of strategies based on sustainability.
Risk Systems That Read Redux
Presenter: Dan diBartolomeo
September 28, 2017
Abstract:
In September 2017 Northfield is introducing the first commercially available factor risk models that incorporate computerized analysis of news text directly into volatility risk forecasts for individual stocks, corporate bonds, industry groups and ETFs based on market indices.
We believe this is the most significant innovation in factor risk models in more than three decades.
Each day the content of thousands of news articles are now part of the input for the full range of "near horizon" models available from Northfield. The line of research that led to this innovation stretches back to 1997 and includes two published papers by Northfield staff diBartolomeo and Warrick (2005) and diBartolomeo, Mitra, Mitra (2009).
In this presentation, we will cover the text analytics that are now part of model inputs and the framework of how the text analysis is used to condition our factor models for both equities and corporate bonds.
The latter half of the presentation will focus on how this model enhancement will contribute to improved portfolio returns by providing a positive definition of security specific risk, as opposed to being the dispersion of the residuals of the factor modeling. To the extent that security specific risk is widely used to represent the opportunity set for active management to add value, improved assessment of security specific risk should add effectiveness to quantitative strategies
Northfield’s Model Blending Technology
Presenter: Mike Knezevich
September 14, 2017
Abstract:
Various vendors provide models with different forecast horizons, normally consisting of short-term, mid-term and long-term models. Each using more or less frequent observation dependent on the forecast horizon: hourly, daily, weekly or monthly and different half-lives to make the model more responsive to recent events for short term investors or longer half-lives for long term investors. As markets change over time, why should investors be limited to a few specific forecast horizons?
Northfield introduced blending in 2013 with the release of our Adaptive Near Horizon Risk models which instead of relying on more frequent data, incorporates contemporaneous market conditions into the existing model structure. Since the Adaptive Near Horizon Risk models have the same factor structure as our long-term models, we can combine the models to create a customized risk forecast horizon between 2-52 weeks depending on the amount of contemporaneous data an investor wants to be represented in their forecast.
Northfield is currently releasing the game changing Risk Systems That Read, which we believe will be the biggest step forward in risk modeling for asset management since the creation of the multi-factor risk model in the 1970s. These models are news conditioned versions of our Adaptive Near Horizon Risk models available in equity as well as multi-asset class models.
Blending combined with the Risk Systems That Read provides investors the ability to incorporate the powerfully accurate risk measurements of news and sentiment into a custom forecast specific to their needs.
Designing Quantitative Strategies
Presenter: Leigh Sneddon
August 22, 2017
Abstract:
This webinar provides a design tool for quantitative strategies. It allows portfolio managers to increase information ratios, to better understand the performance implications of design choices, and to use what-if scenarios to make more informed risk-return tradeoffs.
It includes formulae for active risk, active return, turnover, signal and factor exposures, and other quantities. These design formulae are model-based, compact, and quickly evaluated. Several examples demonstrate their use, for example incorporating smart beta investing, and socially responsible tilts, into active stock-selection, and using custom risk controls. Other examples illuminate the impact of correlations on performance and challenge our understanding of breadth. Monte Carlo simulations verify the design formulae, and qualitative descriptions provide the intuition behind the observed behaviors.
To close the webinar, Leigh will also tell you about a major break-through in attribution for quantitative strategies.
Fixing Active Management: Why Value Investing Works (or At Least Has Worked)
Presenter: Dan diBartolomeo
July 27, 2017
Abstract:
One of the most widely accepted concepts in modern equity investing is that "value oriented" active strategies tend to outperform other active strategies. In many cases these strategies appear to produce a statistically significant risk-adjusted excess return, contrary to the expectations of the Efficient Market Hypothesis. We will argue that this apparent anomaly does not arise because value strategies are innately superior. The way in which equity valuation methods are usedhas been oversimplified to the point of being consistently and materially biased. The major result of these biases is that active management as a whole appears ineffective compared to passive strategies. Value strategies are merely the least impacted among popular thematic approaches (e.g. smart beta, factor tilts, etc.) to equity investing, so they appear superior to other strategies.
We will focus on four aspects of traditional equity valuation methods that are the key sources of the bias. The first issue is the assumption embedded in classical valuation models (e.g. dividend discount models) that companies exist in perpetuity when they obviously do not. Some companies go bankrupt and many others are acquired for minimal value in periods of distress.
The second issue is that projected future cash flows are discounted without regard to the shape of term structure (and the implications of that term structure for forward interest rates). Clearly, the daily operation of bond markets is a clear rebuke to this view.
The third issue is the lack of mathematical distinction between mean arithmetic returns and mean geometric returns which is implicit in the single period of nature asset pricing models (CAPM, APT).
The final issue is that average company level growth projections produced by fundamental analysts routinely imply future levels of national GDP growth that are often multiples of consensus expectation for growth in the economy as a whole. Just as not every child can have above average school grades, it is impossible for the average company to grow faster than the economy as a whole in perpetuity.
Proper consideration of these issues is highly explanatory of the purported value anomaly through history.
Optimizations for Separately Managed Accounts
Presenter: Steve Dyer
July 11, 2017
Abstract:
Separately managed accounts are very attractive to many investors and advisors alike due to the flexibility to customize to investors' individual preferences while being more tax-efficient than a typical mutual fund.
In this 30 minute online workshop, we will show users of the Northfield Optimizer how you can incorporate the many types of preferences and restrictions into an optimization, including capital gain mandates, asset allocations, sleeve restrictions, and cash withdrawals.
Active Mismanagement: Defining Optimal Portfolio Turnover
Presenter: Dan diBartolomeo
June 27, 2017
Abstract:
In the current debate among active, passive and quasi-passive (e.g. smart beta) approaches to portfolio management, there are two concepts of what active means. This term is sometimes used to mean "differences from benchmark" while in other contexts it refers to the frequency of trading of the securities in the portfolio. Studies such as Elton, Gruber and Blake (1996) suggest that active managers exhibit risk-adjusted positive relative returns before transaction costs and expenses, but lag passive indices after costs.
In this presentation, we will assert that active managers trade too frequently as a result of an improper conception of how turnover levels vary across securities within a portfolio. We will present a simple metric that distinguishes turnover levels among different positions, and allows for more appropriate trade-offs between alpha potential and transaction costs. The approach builds on diBartolomeo (2012) which bridges traditional single period optimization with less tractable multi-solutions.
Use of Factor Models in the Presence of Higher Moments
Presenter: Dan diBartolomeo
May 30, 2017
Abstract:
Implicit in the design of traditional linear factor models is the concept that the distribution of individual asset returns is at least symmetric if not Gaussian. This leads many to believe that it is not possible for factor risk models to do a good job of dealing with risk for securities where there is a structural expectation of skew or kurtosis in the asset return distribution, as would commonly arise from many assets such as options or "structured products." As such, there is a belief that the portfolio risk of portfolios containing assets with non-linear behavior must be evaluated through "simulation and repricing."
In this presentation, we will show that repricing is not necessary, or even desirable under the vast majority of cases. We will discuss how the higher moments of the return distribution of individual assets can be sufficiently represented in the context of factor models, and how this information can be combined into portfolio level risk following the methods described in Satchell and Hall (2013). Finally, we will show that our representation of higher moments is sufficient to allow for effective portfolio optimization in the presence of non-zero estimation error of future portfolio returns and portfolio (i.e. the real world).
Total Fund Risk Services
Presenter: Richard Pearce
May 11, 2017
Abstract:
This presentation will provide an overview of Northfield's various total fund risk services and delivery platforms. These are based on the time series derived Everything Everywhere (EE) model that is uniquely well-suited for multi asset class portfolios.
EE covers each investment individually using a single relatively small set of predictive risk factors, so that interrelationships are easily observed and understood. All investable assets are related to the same consistent set of factors. A limited number of factors allows for a more stable model with fluid regime shifts. Fewer factors are also better when trying to decide on meaningful allocations at the total fund level.
Even the most complex investments (derivatives, convertible bonds, privately held real estate, infrastructure, etc.) can be effectively covered using the building blocks of composite assets to model exposures to the parsimonious factor structure.
Retail Mutual Fund (mis)classification Evidence based on Style Analysis V2.0
Presenter: Daniel Mostovoy
April 26, 2017
Abstract:
One of the most difficult aspects retail investor portfolios such as defined contribution retirement plans face is that if the plan is to include actively managed funds, someone must make a decision about which funds or fund families should be considered by investors. Obviously, all investors would prefer fund managers that are believed to be very skilled and who manage risk in a prudent fashion. Regulation of retirement investments in most countries also requires that such considerations be included in making fund choices.
In this presentation we will illustrate Northfield's analytical approach to a particularly troubling aspect of fund risk. One of the key risks of investing in a fund is that the fund is not being managed in a way which is consistent with the stated fund objectives, as first explored in diBartolomeo and Witkowski (FAJ, 1997). We will also revisit our analytical process for properly classifying funds into peer groups, and update the empirical results of the original study. The more recent findings suggest that the classification of funds into groups with purportedly similar objectives remains problematic with many funds pursuing investment strategies which are not congruent to stated objectives. There is a material economic cost to investors as it becomes difficult to construct a diversified portfolio.
Incorporating ESG Considerations into Optimizations
Presenter: Steve Dyer
April 13, 2017
Abstract:
The implementation of environmental, social, and governance (ESG) screens in investment decisions has been steadily increasing over the past decade - around 30% of assets under management in the US by some metrics. And while there is vigorous debate about the relative long-term performance of socially screened universes and ESG metrics, (Northfield research has shown from no significant performance difference to small outperformance of socially screened universes), Northfield has several different ways to incorporate ESG or other metrics or preferences that are not strictly economic into building your optimal portfolio.
This online workshop will show Northfield Optimizer users how to design an optimization with ESG considerations according to your investment strategy, from simple screening procedures in the investment universe to more granular approaches.
Risk, Uncertainty and Time Horizon: What Most Risk Models Get Wrong!
Presenter: Dan diBartolomeo
March 30, 2017
Abstract:
Providers of risk systems often gloss over the most important attribute of any investment risk estimate. It is obvious that all risk of investment performance is in the future. If so, how far in the future: a day, a month, or a century? As we move from the intraday horizons of trading operations to the multiple-decade actuarial horizons of a sovereign wealth fund, the nature of the estimation problem changes profoundly, a fact which risk system providers often prefer to obscure in any effort to present their commercial offering as "one size fits all."
In this presentation, we will address several common errors in the modeling process that can greatly impact the validity of risk forecasts. Among these is the failure to distinguish between statistical risk (i.e. a known distribution) and "Knightian uncertainty" (uncertain distribution subject to change). A second error is the assumed shape of factor and asset return distribution in terms of "fat tails" over shorter horizons. Another frequently observed mistake is to include too many factors in the model than the available data can support in an effort to make risk decomposition reports more granular. Finally, the most common problem is the obvious mathematical error of using a relatively short estimation sample period in order to make a model "more responsive" to changes in volatility levels, while annualizing those same time-varying risk assessments under mathematical assumptions that are valid only if volatility is constant.
Minimum Variance Portfolios
Presenter: Jason MacQueen
February 28, 2017
Abstract:
There are three basic theoretical ways of building multi-factor risk models for equity markets: by running cross-sectional regressions, given the stock betas, to estimate the factor returns in each period; by running time series regressions, given the factor returns, to estimate the betas for each stock, and statistically, by applying principal component analysis to a stock covariance matrix.
In practice, modern risk models are often hybrids, using various combinations of the three methods outlined above. At our St Petersburg Conference in November 2015, we presented 'Horses for Courses', a comparative analysis of the Northfield US Fundamental risk model (a cross-sectional model), the Northfield US Macro-Economic hybrid risk model (APT-based), and the Northfield US XRD hybrid risk model (primarily a time series model).
We now consider a different kind of test, by using each of these models to manage Minimum Variance Portfolios over an 11-year period, rebalanced quarterly. Since Minimum Risk portfolios pay no attention to expected returns, this represents a fairly pure test of a risk model. To make the exercise a little more interesting, we have also included a pure Statistical model as well. The results will shed further light on how well the models perform in four different periods: the Low Risk Bull Market from January 2005 to July 2007, the GFC from August 2007 to March 2009, the Recovery from April 2009 to December 2012, and the New Normal, from January 2013 to November.
Risk Tolerance in Optimizations
Presenter: Steve Dyer
February 9, 2017
Abstract:
For math-oriented finance practitioners and quants especially, determining your own or a client's risk tolerance can be the most difficult part of building a portfolio since it relies on hard to quantify feelings, fear, expectations management, and other difficult-to-regress concepts.
The risk tolerance is also one of the most important, yet most frequently misspecified inputs in an optimization.
This workshop is designed for users of the Northfield Optimizer to help them better understand the risk tolerance parameter, best practices and frameworks for determining their risk tolerance, and common missteps that can go undetected and how to avoid them.
Replicating Residential Real Estate Using Liquid Market Securities - A Case for Risk Factor Models for Everything Everywhere
Presenter: Rick Gold and Emilian Belev
January 26, 2017
Abstract:
The value of the U.S. residential real estate stock is estimated to be tens of trillions of dollars making it perhaps the nation's largest investable asset class. While the vast majority of homes are not held for investment purposes, as home prices begin to wake up from their post crisis slumber, institutional investors are taking a renewed interest in this asset class as they search for yield. Conversely, mortgage lenders have made an implicit bet on home values as collateral for their outstanding loans. Homeowners themselves should be interested in the ability to hedge the risks of homeownership since their home equity typically represent a significant percentage of their net worth. Lastly, those seeking some of the benefits of homeownership (e.g. prospective home buyers saving for a down payment) should be interested in creating synthetic portfolios which mimic the movement in home prices. All of these observations point to the need to reliably measure the risks of residential real estate and to construct effective hedges using liquid, divisible, and accessible investment choices.
In this presentation we outline our approach to creating a replicating portfolio for some of the most widely followed residential real estate benchmarks such as the S&P CoreLogic Case Shiller house price index and the Federal Housing Administration HPI while simultaneously following the implications for the underlying risk factor model. We pay special attention to the question of smoothing, normally inherent in real estate indices, and how to account for its effect.
The proposed methodology is in every way analogous to the approach investors would use when creating hedges for commercial real estate equity and debt. Since real estate and property taxes are, respectively, the backbones of mortgage trusts and municipal finances, this modelling technique is also a key tool in decoding the risks of MBS and general municipal bond obligations. Both asset classes represent several trillion dollar markets in their own right.
Finally, due to the common risk factor structure, this approach is an essential tool for risk management and portfolio construction at the enterprise portfolio level.
Risk Parity, Factor Investing and US DOL Regulation of Defined Contribution Retirement Plans
Presenter: Dan diBartolomeo
December 20, 2016
Abstract:
Regulators in numerous countries (USA, UK, Switzerland, etc.) have strengthened the fiduciary obligations on providers of financial services, whether they be an individual financial advisor, or a large asset management firm offering hundreds of mutual funds. One of the most interesting requirements has been put forth by the US Department of Labor as part of the regulation of defined contribution retirement plans regarding the distinction between "open and closed platforms." For example, if a mutual fund group provides administration of defined contribution retirement accounts (e.g. 401K plans), it must either allow investors to invest in outside funds that are not managed by the fund group, or be able to demonstrate that the range of internally managed funds offered is sufficient to meet all the needs of their investor clients. We believe this issue has related implications for asset managers in many countries.
Northfield has been engaged by fund groups to assess their preparedness with respect to this regulatory issue. In carrying out these analyses, we have determined that two ideas play a critical role. The first concept is asset allocation by risk parity. For example, the return properties of a portfolio consisting of an investment in a long maturity bond fund, and a money market fund can be closely replicated by holding a larger quantity of an intermediate maturity bond fund, and a smaller position in cash. To the extent that these sorts of common properties exist across asset classes, we can also describe them in terms of a common factor structure as described in diBartolomeo (2015).
In this presentation, we will both illustrate the analytical methods we have employed to make determinations of "sufficiency" of a fund family to meet needs of investors as defined by the DOL regulation, and provide an intuitive understanding of the outcomes.
Equity Factor Timing and Kiddie Bowling
Presenter: Dan diBartolomeo
November 29, 2016
Abstract:
Relatively few equity investment strategies try to generate alpha by "timing" factor returns by forecasting period by period returns to equity factors. In such a strategy, any factor which explains security covariance may potentially be useful, as opposed to the traditional approach to generating abnormal returns by creating exposure to risk premia (factor returns with persistent positive mean). Successful factor timing strategies may arise from factors with zero means, but with predictable serial properties.
In this presentation we will explore a new and promising approach to factor timing. The methodology assumes that factor returns are positively serially correlated under normal conditions (factor trends) but where the first order autocorrelation coefficient will change sign as cumulative factor returns become large. A familiar example of this structure would be discount rates set by central banks which are frequently subject to long series of changes of same sign, but without any observation of non-stationarity which would be expected to eventually result in extreme values. The process can be visualized like a bowling ball zig-zagging down the alley by bouncing off the "bumpers" provided to children. The vast majority of observations represent a simple directional trend, but at the extremes the trend inevitably reverses.
We will show empirical examples of multiple Northfield model factors that appear well suited for alpha generation.
Advanced Techniques for Wealth Managers and Family Offices
Presenter: Dan diBartolomeo
October 27, 2016
Abstract:
Dealing with the investment management needs of super affluent households brings different challenges as compared to the typical retail investor households or even the "mass" affluent.
In this webinar we will focus on four key areas of investment practice that illustrate the old adage that the "rich are different." The first will be the handling of large concentrated positions in legacy portfolios. Often, the founding families that have started successful firms are left with huge portions of their wealth locked into a low-cost basis position in a firm they no longer control. We will illustrate the "complementarity portfolio" approach to resolving this problem. Secondly, we will look at optimal spending policies, and illustrate a simple technique for ensuring that investment policies (particularly asset allocation) and spending policies are properly coordinated. Third, we will analytically examine the reasons why many ultra-wealth families are very conservative investors, despite the fact that they could afford to take on more aggressive investments to improve returns. Finally, we will examine the concept of "householding" in which an optimal portfolio is formulated for an aggregate set of family members, rather than for each family member individually.
Contrary to simplistic forms of this process being marketed by some asset managers, our conception of the problem allows for both differential tax circumstances and different levels of risk tolerance across family members. In addition, our process allows the family to incorporate "fairness" considerations into the process. The goal is to ensure that none of the individual portfolios legally held by each family member is particularly sub-optimal in order to enable optimality for the aggregate.
Credit Spreads – The Flipside of the Default Distribution
Presenter: Dan diBartolomeo
September 27, 2016
Abstract:
Bond investing has consistently used credit spreads as a cornerstone of determining payoffs and risk. A more recent work in structural credit risk models establish a connection between the assets of an enterprise, its equity claim, and credit viability. These two schools of thought largely avoided the need to reconcile their beliefs in the absence of common language.
Since the introduction of the structural credit approach to our fixed income risk model in 2011, Northfield has made steps to write the syntax of this language. It showed that the two frameworks merge into one when put in a general framework. That is the utility of a rational investor relates changes in spreads to higher distributional moments both for reasons of multi-periodicity, as well as the projected default payoff distribution.
We offer our empirical work to show how this theory holds in practice.
Passive Management, Market Efficiency and Long Term Return Premia
Presenter: Dan diBartolomeo
August 25, 2016
Abstract:
The recent surge in passive management has largely arisen from an increasing portion of investors coming to the conclusion that active managers cannot outperform the market net of costs. One example of the evidence for this trend is the increasing use of factor tilted passive strategies by large asset owners such as sovereign wealth funds. The theoretical rationale for such policies rely on the concept of market efficiency. Traditional concepts of an efficient market rest on the requirement that asset price changes must be largely unpredictable, and thus it is therefore quite difficult to obtain abnormal risk-adjusted returns. We suggest that for long-term investors a different definition of market efficiency is more relevant. An efficient market is one in which investors with heterogeneous preferences trade amongst themselves to set asset prices such that the global uncompensated disutility of not obtaining preferred investment characteristics is minimized.
This presentation will describe numeric programming solutions for the global optimal portfolio and also methods for defining investor preferences. With these two building blocks, we can derive the risk premia associated with various asset classes as shadow prices. We will also demonstrate how long term demographic trends must foretell the shifting of investor preferences, and thereby predict changes over time in the return premia, which compensate investors for unresolved disutility.
Estimation of Event Risk: Crashes, Brexit and Whatever Comes Next
Presenter: Dan diBartolomeo
July 26. 2016
Abstract:
The recent "Brexit" vote has created massive anxiety among investors in the UK, with lesser but still severe concerns across the European Union countries. One outcome of particular note for investors is the huge outflows from UK property funds forcing the funds to curtail withdrawals. While numerous financial firms have circulated commentaries on this subject from their investment strategists, almost all of this discussion is qualitative interpretation of the known facts. A few have offered highly speculative risk discussions based on hypothetical stress tests with minimal foundation in either economic theory or empirical fact. We assert that most systems routinely used to assess the risk of extreme events and other short lived phenomena (e.g. commercial providers of VaR and CVaR measures) are simply inadequate, and often lead to counterproductive actions by long term investors.
The body of the presentation will address numerous key aspects of how event risks should be viewed, assessed and incorporated into portfolio management actions. Among those key issues are the use contemporaneously observable market information to improve risk estimates, adjustments in the assumed distribution of asset returns, inclusion of serial properties of asset returns, liquidity analysis and scenario driven conditioning of volatility estimates.
Global Wealth Management and the Panama Papers
Presenter: Dan diBartolomeo
June 30, 2016
Abstract:
In 2010, I was among multiple speakers at a large CFA wealth management conference in Singapore that warned of the financial services equivalent of a "time bomb." The thesis was that the globally common practice of wealthy individuals evading taxes in their home countries by diverting money into offshore shell corporations would eventually blow up. With the recent public release of thousands of documents hacked from Panamanian law firm Mossack Fonseca, the first explosion has been realized, with substantial negative publicity for both financial firms and myriads of high profile clients.
In this presentation, we will first review common practices in terms of offshore corporations both for individual private investors, and for collective vehicles such as hedge funds. While heavily tainted with the stain of tax evasion there are other legitimate benefits of such vehicles. The wealth management industry in some countries such as the USA and Australia have made extensive progress toward broad availability of tax sensitive investing. Even Harry Markowitz, Nobel prize winner for pioneering work in portfolio theory recently published research work on "tax cognizant" portfolio allocation (with Kenneth Blay, Journal of Investment Management, 2016). Unfortunately wealth management organizations in the rest of the world have remained in denial, choosing continued reliance on secretive offshore vehicles to tread the thin line between legal tax avoidance and illegal tax evasion.
The presentation will conclude with an overview of how the techniques of tax sensitive investing can be adopted quickly by non-US financial firms so as to prosper in the "post Panama papers" world of wealth management.
Making Lifetime Investing Planning a Reality
Presenter: Dan diBartolomeo
May 24, 2016
Abstract:
While almost all wealth management organizations talk about the concept of an customized investment plan for each client household, essentially none deliver on that promise. A plan is a set of contemplated actions for the future. Instead, wealth managers provide recommendations only for the current asset allocation of the investment portfolio with the only "plan" being to revisit the allocation in a year or two. Among the financial products available to retail investors are target date funds that include the concept of an allocation glide path, but such funds are based solely on expected year of retirement. They are not sensitive to wealth levels, non-retirement financial goals or the potentially complex preferences of high net worth investors.
In this presentation, we illustrate a process to create the "maximum likelihood" forward time series of expected asset allocations through the investor's lifetime (now, next year, 2 years out, 5 years out, etc.) using the life balance sheet concept described in Wilcox (2003), the non-parametric preference functions from Bolster and Warrick (2008), and a process to combine these two disparate concepts from diBartolomeo (2014). The delivery of an actual investment plan reassures investors psychologically as they can see life events (e.g. college expenditure) reflected in the planned changes in asset allocation. In addition, the conditional foreknowledge of "what we are doing next" allows much of portfolio rebalancing to be done through cash flows (savings inward, reinvestment of income, and spending outward) thereby reducing transaction costs and often large taxes associated with periodic rebalancing. Having an actual plan also contributes to reduced uncertainty as to where to liquidate assets within the portfolio to augment investment income in today's low yield environment.
Custom Hybrid Risk Models
Presenter: Jason MacQueen
April 28, 2016
Abstract:
All standard equity risk models can be used to estimate portfolio risk and tracking error, to show the beta of a portfolio to its benchmark, and to give a decomposition of risk into factor-related and stock specific parts. Most standard risk models can also be used to provide a risk decomposition showing the contribution of each stock held in the portfolio to the overall risk or tracking error.
However, many active managers also want to know about the various factor bets in their portfolios, and here standard risk models are not always quite so useful. The factor risk decomposition will always be expressed in terms of the factors used to build the risk model. All too often, these factors do not correspond to the factors the manager had in mind when the portfolio was built.
This mismatch may be as simple as the risk model using broad Sector factors, while the manager was thinking in terms of Industry bets, or it may be that the manager's definition of a Style factor, such as Value, doesn't correspond to the definition used in the risk model. Perhaps the risk model factors cover both Emerging and Developed markets, while the manager is focused on just a Developed Europe portfolio, or perhaps the factor has been orthogonalised on the market, while the manager is betting on the factor in its natural state.
Unless the factors in the risk model correspond to the factors used to select stocks and rebalance the portfolio, the manager will not be able to identify and quantify the factor bets clearly. To do this, the manager needs a Customized Hybrid Risk Model (CHRM) which mirrors the investment process used to manage the portfolio.
Everything about a CHRM can be customized: the universe of stocks, the industry classification, the country groupings (in a global or regional model), the treatment of currency and style factors, the periodicity, time-weighting, investment horizon and so on. All the CHRMs we build at Northfield are Hybrid models, which means that rather than assuming the chosen set of factors will necessarily capture all the systematic covariance in the universe of securities, we always add a small number of statistical factors to the model to capture any residual systematic covariance, or any new factor effects that may arise in the future.
This presentation will discuss the benefits of using a Custom Risk Model in detail, and provide a number of examples of CHRMs built for different clients.
Diversification and Real Estate, Part II
Presenters: Rick Gold and Emilian Belev
March 29, 2016
Abstract:
In previous webinars, we focused our efforts on analyzing the correlation between real estate and other major asset classes. In particular, we examined the effects of appraisal bias on traditional correlation techniques and proposed methods to correct for these effects in order to extract private equity real estate's true inter-asset class correlations. The results revealed an unequivocal and statistically significant link between real estate and both stocks and bonds.
In this presentation, we will demonstrate the practical implications of our study to an investor pursuing an efficient diversification strategy. Specifically, we will employ a risk modeling approach with a robust historically observable relationship to both direct (illiquid) and securitized (REIT) real estate investments. To our knowledge, this is the first time that a strong empirical connection between these two asset classes can be demonstrated, despite the logical and practical desire to find such a link. The finding is even more noteworthy in view of the fact that the risk model used to connect the two asset classes is based only on real estate's fundamental characteristics, and neither commercial real estate indices and/or REIT returns were used as inputs.
We will also discuss the implications for hedging real estate from the perspective of a fund investor whose payoffs are based on appraised values (an open-ended real estate fund), as well as an investor whose payoff depends on the arm's length transaction of the underlying property itself.
Rules-Based Style Rotation: Dynamic Switching between Smart Portfolios
Presenter: Jason MacQueen
February 23, 2016
Abstract:
Smart Beta has become the latest fashion to conquer the investment community. From a quant perspective, these are simply factor portfolios which offer significant exposure to the desired Style factor (subject, of course, to the usual long-only restriction).
However, the weighting schemes of most Smart Beta ETFs make no attempt to trade off expected return against risk, or to minimize their exposure to unwanted factors, with the result that whatever style tilts they do have are likely to have only a modest effect on their performance. It should be, and is, easy to create far more efficient Smart Beta ETFs by using standard portfolio construction principles. These are called Smart Portfolios.
The underlying rationale for such products is that Style factors exhibit reasonably persistent risk premia. Thus, both Value and Momentum Smart Portfolios offer the prospect of higher risk-adjusted returns than the market, on average over time, although both will also suffer periodic underperformance.
In this talk we consider a dynamic strategy of switching between a set of Smart Portfolio ETFs, each capturing the returns to an individual Style. We focus on identifying the Smart Portfolio with the most consistent performance over the in-sample period, which we take to be a measure of the persistence of the corresponding Style factor risk premium. While this may not be the one with the highest returns, it is more likely to perform reasonably well in the next, out-of-sample period. We illustrate this dynamic strategy over the past 10 years in the US market.
An Optimized Approach to Scenario Driven Risk Simulations
Presenter: Dan diBartolomeo
January 28, 2016
Abstract:
This presentation provides a new approach to risk assessment from numerical simulations. As risk-related regulation extends from commercial banking to other parts of the financial services industry, risk assessments arising from "stress tests" and "scenario analysis" have become more widely discussed and implemented. Unfortunately, traditional methods for this kind of risk assessment are often counter-productive for long term investors who are not levered, as described in two past Northfield newsletters (May and September 2006).
Existing processes have worked in one of two ways. The first is Monte Carlo simulations where there is random sampling from a parametric or empirical distribution to get a range of possible outcomes. Risk assessment is based on the lower tail of the portfolio value distribution. The second process is to forecast a single return value for a series of specific exogenous scenarios. For example "What will be the % change in the value of my portfolio if interest rates go up 2%? and oil prices go down 30%." It is argued that if we look at enough different "stress scenarios" we can gain an intuition about "worst case" outcomes. Unfortunately, the way most stress scenarios are formulated, their actual probability of occurrence is very, very small. Investors predicating investment strategy on such low probability outcomes end up with portfolios that are materially sub-optimal in the vast preponderance of situations.
To resolve the shortcomings of numerical methods we have built a new process, extending the approach suggested in Meucci (2008) which combines Monte Carlos simulations with the flexibility to overlay complex explicit scenarios. The computational process involves an optimization problem that calibrates our "bootstrap" resampling process (see our newsletter June 2013) to one or more user defined scenarios. The analytical output of the process is a robust representation of the distribution of possible outcomes, while being consistent with any mathematically feasible "stress scenario."
Reconciliation of Default Risk and Spread Risk in Fixed Income
Presenter: Dan diBartolomeo
December 29, 2015
Abstract:
There are two conflicting concepts of what credit risk actually is. The classic definition has to do with the likelihood that a given fixed income instrument will default (Probability of Default, PD), and the expected severity of economic loss in the event of a default (Loss Given Default, or LGD). In this view, the focus is on the "tail risk" (negative skew in the return distribution) associated with the default event. Many fixed income market participants prefer to think of a given fixed income instrument as offering a credit related yield spread above a comparable duration riskless instrument. These investors think of credit risk as the volatility of the credit yield spread and related impact on the market value of an instrument (conditional on the duration). If investors are not risk-neutral, the credit spread will compensate investors for their expected loss (PD*LGD), plus provide a risk premium to induce risk averse investors to hold these instruments.
These two concepts of credit risk are not equivalent because credit spreads can change over time both because of changes in expected loss, and separately because aggregate investor risk aversion can change, forcing a change in the risk premium (incremental yield) which fixed income borrowers must pay.
In this presentation, we will review relevant approaches to credit risk, and illustrate how to reconcile the two views in order to satisfy the default risk concerns of "buy and hold" investors, while simultaneously explaining yield spread volatility for investors who are more concerned with controlling variation in period to period returns.
Risk Model Testing, or Horses for Courses
Presenter: Jason MacQueen
November 24, 2015
Abstract:
The decision as to which risk model an asset owner or a fund manager should use balances a preference for a particular view of the world (such as fundamental, macro-economic, or customized) with risk model accuracy in both boom and bust phases of the economic cycle.
This talk will present the results of testing several different risk models over the last 10 years, including the GFC, and discuss the market conditions and types of portfolios for which the different models performed better or worse.
We will also cover details of risk model construction, the use (or not) of dummy variables, and the use of Bayesian statistics for determining whether a security is exposed to which individual factors.
Back-testing: A Useful Tool or "Financial Charlatanism"?
Presenter: Dan diBartolomeo
October 22, 2015
Abstract:
Back-testing is the widely used practice of simulating an algorithmic investment strategy. While essentially everyone involved in quantitatively driven investment methods conducts back-tests, it is widely accepted that simulated investment results achieved "in sample" are at best only a very weak indication of results to be expected in "out of sample" experience.
In this presentation, we will describe the causes for the minimal validity of back-tests, and suggest methods to mitigate the problems. We will discuss the current day implications of material from seminal studies by Kahn and Rudd (1995) on the relationship of past and future performance, Kahn (1997) on common statistical errors in investment tests, and diBartolomeo (1999) on the conceptual and philosophical limitations of back-tests. The final portion of the presentation will be devoted to a detailed exposition of how practitioners can limit the risk of "overfitting", based on the mathematical framework of Bailey, Borwein, de Prado and Zhu (2014).
Behavioral Aspects of Risk
Presenter: Dan diBartolomeo
September 29, 2015
Abstract:
Since the early work of Daniel Bernoulli in 1740, it has been widely acknowledged that investors generally do not like risk, but they are largely ineffective at describing how financial risk should actually be defined.
In this presentation, we will begin with a philosophical and semantic discussion of what risk is and how investors talk about it. We will then move on to a review of plausible investor utility functions so as to have a context from which to distinguish what seems to be very sensible behavior by investors in response to investment risk, from apparently irrational behavior.
The remainder of the presentation will focus on high level behavioral aspects of risk, and how the many seemingly bizarre behaviors arise from investors and managers trying to give the appearance of good risk management, as opposed to the reality of good risk management. Once this distinction is clear, we see that the attitudes which drive investor behavior regarding risk often range from "willful ignorance" to "delusion." Many regulatory schemes around the world also reinforce the irrational behavior, as risk regulations intended for commercial banking have been poorly revised for asset managers and asset owners.
Risk Model Testing and Regulatory Reporting
Presenter: Dan diBartolomeo
August 27, 2015
Abstract:
Increasing regulation of non-bank financial institutions such as asset managers and pension funds has brought about many requests from our clients in regard to how to test the efficiency of the risk models they use, and how to report this information to regulatory agencies. In this presentation, we will describe several different methods for evaluating the predictive power of risk models over different time horizons in the context of "model risk" as described in US Federal Reserve SR 11-7 and related UCITS requirements. In addition, we will discuss the conflict between a focus on risk reporting as a matter of regulatory compliance and the practical considerations of managing a long horizon, unlevered investment fund. In an increasing number of organizations this distinction has led to frequent conflicts between two risk systems, one used for portfolio decision making and a separate system for compliance reporting.
Diversification and Real Estate, Part I
Presenters: Rick Gold and Emilian Belev
July 28, 2015
Abstract:
Private equity real estate has traditionally been touted as an asset class with superior diversification qualities inasmuch as it has been shown to have low volatility and correlation with other asset classes. Unfortunately, the asset class' reliance on appraisal-based indexes has not only served to propagate this myth, but it has also kept investors in the dark with respect to the asset class' true behavior and its relationship relative to not only other investment vehicles but other real estate investments as well.
In the first half of this discussion we will use data from two index providers and show how it relates to the performance of other asset classes. We will demonstrate that real estate diversification vis-à-vis other asset classes is mostly a byproduct of computing historical correlations without accounting for regime changes. In the second half of this presentation we will examine what are the implications of our findings and show that real estate is less of a diversifier than claimed by index-based methodology proponents. Specifically, we will explore how to manage and hedge real estate risk arising at the total portfolio level using Northfield's enterprise level risk model including both public and private asset classes.
Assessment of Corporate Credit and Counterparty Risk Using News Flow and Sentiment
Presenter: Dan diBartolomeo
June 30, 2015
Abstract:
Since the Global Financial Crisis of 2007-2009, history has been marked with numerous failings to correctly assess the credit worthiness of financial instruments, financial institutions and governments. Institutional confidence in the traditional credit rating agencies has been greatly reduced. One of the largest rating agencies, Standard and Poor's, recently agreed to pay a $1.4 billion fine to US regulators for alleged widespread negligence in the ratings of certain complex financial instruments.
As an alternative to the traditional rating process, this work will illustrate Northfield's proprietary RISK SYSTEMS THAT READ process of news flow and sentiment statistics to calibrate and update the credit risk of corporations and financial institutions in real time. A modified version of the Merton (1974) contingent claims model from diBartolomeo (2010, 2012) is used to break each corporate debt into two pieces, the first considered riskless debt and the second portion considered to be equity in the issuer. We utilize news flows and sentiment statistics to frequently update the expected volatility of the assets of the firm and hence the credit risk of the debt in terms of both the probability of default and loss given default.
Risk Systems That Read
Presenter: Dan diBartolomeo
May 28, 2015
Abstract:
Analytical models in finance all share some basic concepts. Financial market participants observe some period of past events they deem relevant, build a statistical model of the observed data, and then make the heroic assumption that events in the future will be like those in the past. While almost every financial institution has extensive risk modeling systems in place (as often mandated by regulators) the Global Financial Crisis has shown that such systems are frequently grossly inadequate. What is missing from nearly all models is a recognition of how the present is different from the past, and therefore the short term future is also likely to be different from the past. By defining "news" explicitly as the information set that informs us of the differences between past and present, we can condition our estimates of the distribution of future outcomes more robustly.
Building upon the methods in diBartolomeo, Mitra, and Mitra (2009), and Kyle, Obizhaeva, Sinha and Tuzun (2012), we will introduce a new approach to using quantified news flows and related sentiment scores in the prediction of asset portfolio risk. This new process can operate in real time, and can address tens of thousands of global companies and financial institutions (for counterparty risk).
The Choice of Model Factors Under Multiple Definitions of Risk
Presenter: Dan diBartolomeo
April 30, 2015
Abstract
Risk assessment in asset management is intrinsically multi-dimensional. Investors may be concerned in differing degrees with a multitude of risk measures such as tracking error, active risk, absolute volatility, VaR, CVaR, and "first passage" risk (drawdowns). We will begin with a review of the relevant literature such as Roll (1992) and Wilcox (1994) and Kritzman and Rich (2002). The presentation will then examine how the different definitions of risk might influence our choice of which factors include in both alpha and risk models. If I hold a benchmark index fund and add a low beta stock to it, tracking error will increase but absolute risk decreases. Some factors such as balance sheet gearing (debt/equity or similar ratio) would have comparable behavior. If my portfolio has an average gearing level greater than X, this will increase tracking error and increase absolute risk, while portfolio gearing below X will increase tracking error but must reduce absolute risk (a company with no debt cannot go bankrupt). However, for many factors (e.g. momentum, book/price) there is no immediate intuition as to which way factor bets normally contribute to absolute risk, potentially making factors bets harder to interpret.
The presentation will conclude with a numerical simulation that illustrates the multi-period problem for long term investors. For the normal IID (random walk) almost all risk measures are a scalar of a standard deviation around a mean which itself is unknown. In the long run, we demonstrate that "uncertainty of the mean" is at least as important as "volatility around the mean." We show that it is often helpful to take bets on "alpha" factors that have zero factor return on average if the factor returns are negatively correlated with the market and hence decrease the absolute volatility and "tail risk" of the portfolio.
Optimal Deal Flow for Illiquid Assets
Presenters: Emilian Belev and Rick Gold
March 31, 2015
Abstract
Modern portfolio theory has largely avoided the question of what to do with illiquid assets. In large part, this is not surprising since as their name implies, illiquids operate under different conditions than that of their liquid cousins. Appraisal-based, rather than auction-based pricing, large lumpy assets, and sales cycles often measured in months, rather than milliseconds, are just a few of the differences between the asset classes and also some of the reasons why illiquids have not been the darling of academics and also not easy to fit in standard models. This in turn has made it more difficult for owners of illiquid assets to directly address the fundamental issue facing all investors: what to buy, when to buy, and finally when to sell.
Recognizing that owners of illiquid assets cannot take the same path as their stock and bond counterparts, Northfield has developed a solution which merges techniques from fundamental and quantitative finance to tackle this problem in a unique but sensible manner. By fully integrating illiquids into the same pantheon as traditional holdings, investors can now concentrate their efforts on how to maximize their risk-adjusted returns rather than mulling over the best way to simply calculate their risk-adjusted returns.
“Guaranteed” Alpha: Using Risk Budgeting to Improve Performance by Reducing Management Fees and Other Expenses
Presenter: Dan diBartolomeo
February 24, 2015
Abstract
The one way in which an institutional asset owner can be assured of improving the financial performance of their portfolio is to reduce the expenses associated with operating their fund. The largest of these expenses will be fees paid to agent asset managers. Given popular mixtures of active and passive management across both traditional and alternative asset classes, we estimate that the typical asset owner with $10 Billion AUM spends just under $50 Million per annum in manager fees, much of which is wasted through demonstrably inefficient capital allocations between active and passive management, and among agent active managers. This inefficiency arises because while an asset owner can diversify risks by investing across multiple managers, a manager cannot diversify (or hedge) the risk of their own active performance being poor. The magnitude of management fees is often 100 or 200 times the cost of a sophisticated risk management process that could contribute to large reductions in costs.
The presentation will begin with the theoretical development of the metric “covenant information ratio.” Simply put, how much active risk should a manager be taking based on their own representation of their ex-ante IR, and fee levels. We will then cover two different approaches to resolve the inherent conflict of risk preferences between asset owners and agent managers. The first is a unique variation on the concept of risk budgeting which optimally determines asset allocation, the allocation between active and passive management, and a mechanism to determine optimal aggressiveness levels for active managers. Our second approach is to revisit the concept of Centralized Portfolio Management (diBartolomeo, 1992 and 1999), which has been successfully implemented in recent years. Under the CPM process, management expenses and trading costs are reduced, and greater tax efficiency can be achieved for taxable investors such as family offices. In this presentation, we will first examine whether variation of estimates of asset specific risk across models is likely to be statistically significant or economically material. We will then consider a positive definition of specific risk at both the firm and individual security level based on imposing a no-arbitrage condition on the capital structure of a firm. Once we have a prescriptive estimate of specific risk, we will conclude with a discussion of how conditioning the estimates on alternative information sources such as quantification of text news reports can be used to capture time series variation in the true, but unobservable level of asset specific risk.
Alpha Estimation for Quantitative Asset Managers and the Definition of Asset Specific Risk
Presenter: Dan diBartolomeo
January 7, 2015
Abstract
Investment models routinely make distinctions between factor and idiosyncratic (asset specific) risk. This division is enshrined in theories such as the CAPM and the APT. The estimated magnitudes of stock specific risks are also a key metric of the opportunity set for active equity managers, and are widely used in the scaling of alpha expectations (See Grinold 1994). In the conventional process of constructing factor risk models, we arrive at an estimate of idiosyncratic risk for stocks by virtue of a negative rather than positive specification. We take idiosyncratic risk of an asset to be closely related to the residual portion of the asset's observed return variance that we cannot explain by virtue of our specification of factors rather than actually trying to directly estimate what the true degree of idiosyncratic risk actually is. Such conventional processes have numerous implications that should be of interest to investors. For example, it implies that different factor specifications of risk models may arrive at different estimates of asset specific risk even with the same input data.
In this presentation, we will first examine whether variation of estimates of asset specific risk across models is likely to be statistically significant or economically material. We will then consider a positive definition of specific risk at both the firm and individual security level based on imposing a no-arbitrage condition on the capital structure of a firm. Once we have a prescriptive estimate of specific risk, we will conclude with a discussion of how conditioning the estimates on alternative information sources such as quantification of text news reports can be used to capture time series variation in the true, but unobservable level of asset specific risk.
Smart Portfolios
Presenter: Jason MacQueen
November 20, 2014
Abstract
Smart Beta has become the latest fashion to conquer the investment community. There are now numerous indices and ETFs purporting to offer exposure to one or more investment styles, both in the USA and elsewhere. Its cheerleaders claim that Smart Beta investment products offer the alpha promise of active managers, without the corresponding drag on performance from fees.
Large pension funds or endowments with several active managers will almost invariably find that their diversified equity portfolio, in aggregate, consists of a bet on the equity market plus a relatively small number of style factor bets, so that their fund's performance will be that of the market, overlaid with various style tilts, and minus the managers' fees. Put this way, it is easy to see the appeal of Smart Beta products.
However, there are also critics of both the underlying concept and its many implementations. From a quant perspective, these are simply factor portfolios which offer significant exposure to the desired style factor, but the way in which many Smart Beta products are designed suggest that the style tilt will only have a modest effect on their performance. To suppose that a market-capitalisation-weighted portfolio of US stocks with high Book-to-Price ratios will provide a meaningful exposure to the Value premium is naïve at best, and weighting by their "relative style attractiveness" is little better. Having a simple story to sell Smart Beta products comes at the price of leaving a lot of the Style premium on the table.
This talk will argue that the real added value to be gained from creating Smart Beta portfolios lies in the methodology used to create the Smart Portfolio, and that this, in turn, is the end result of taking care at each step of the portfolio construction process. The talk will cover several US Style strategies, each based on standard Style factors.
Measuring Skill in Active Managers
Presenter: Dan diBartolomeo
October 1, 2014
Abstract
Every active manager and every investor who invests with an active manager must believe that the manager must possess an above average ability to make investment decisions else it would be more beneficial to pursue passive strategies. However, the effective identification of skill (as opposed to luck) in investment management has proven elusive, as the volatility of financial markets creates a low “signal to noise ratio” setting.
In this webinar we will present the Northfield PWER methodology, a sophisticated multiple step process for statistical analysis of manager skill that we began to develop in 2006. Within this method are approaches to creating effective peer groups, finding optimal evaluation periods, weighting past return observations, and a Bayesian construct for the consideration of luck versus skill. PWER ratings of more than 100,000 mutual funds, hedge funds and separate accounts are now conducted on a monthly basis.
To conclude, we will also show how “fund of fund” and multi-manager funds can augment PWER scores with the concept of the Effective Information Coefficient, which uses risk models and position information to simultaneously evaluate both the predictive power and the portfolio construction skills of active managers.
Risk Management Priorities for Asset Owners: What Senior Management and Trustees Need to Know
Presenter: Dan diBartolomeo
August 26, 2014
Abstract
In the aftermath of the Global Financial Crisis, risk management has become a mantra among large asset owners. Unfortunately, much of this attention is misguided and ineffective because it confuses the appearance of risk management (e.g. scheduling frequent committee meetings) and the reality of responsible risk management. In this presentation, we will lay out a prioritized list of steps for effective risk management at the “C-Suite” and board level. While most of the material to be presented will be suitable for all asset owners, we will also spend time to differentiate the needs of defined benefit pension funds, defined contribution funds, endowments, family offices and public mutual funds. Among the topics to be considered are the levels of risk associated with strategic and tactical asset allocation, hedge funds, illiquid investments (e.g. real estate and private equity), and active management. We will also spend time considering the issue of time horizon as most asset owners profess to have a “long term view” but have only a vague notion of the operational aspects of an explicit time horizon for risk. The horizon issue is further complicated in some countries where very short horizon measures suitable for commercial banks (e.g. 1 day VaR) have been incorporated in regulations. Much of the content for this presentation has been incorporated into training materials by the Society of Actuaries, and many of the concepts will be illustrated with sample output from our RAMP risk consulting service package.
Portfolio Optimization with VaR, CVaR, Skew and Kurtosis
Presenter: Dan diBartolomeo
July 16, 2014
Abstract
Since the theoretical advent of mean-variance, portfolio optimization in the 1950s there has been an ongoing debate as to the necessity of including higher moments of return distributions (skew and kurtosis) into the process. In recent years, the increasing regulatory focus on downside risk measures such as VaR and CVaR has extended the interest in this topic. This presentation will first identify the portfolio situations where the importance of higher moments appears to be economically material and statistically significant. We will then use the example of "catastrophe" bonds to illustrate two broad approaches to incorporate higher moments into portfolio optimization. In the first approach, we will review "full scale" optimization methods (including random trial and error) that explicitly include skew and kurtosis in the objective function. In the second approach, we will consider using analytical techniques to reduce the four-moment problem to an approximately equivalent mean-variance problem, before solving conventionally. Due to the estimation error in the parameters, we find transforming the problem into mean-variance equivalence to be sufficient for all but the most extreme cases.
Unlisted Assets in the Context of Enterprise Risk Management
Presenter: Emilian Belev
May 14, 2014
Abstract
With the maturity of Northfield's private equity real estate model into a full blown global model, we have turned our attention in the unlisted asset space to private equity and infrastructure investments employing a similarly intuitive model framework. This presentation starts with an introduction to the model's theoretical underpinnings which in a familiar fashion disassembles these complex investments into their fundamental building blocks, analyzes them, and then intuitively re-assembles them into a coherent whole, using a global intra-asset class risk model. We shall also point out key similarities and differences with our real estate methodology along the way. We will also compare and contrast Northfield's methodology with other popular alternatives. Following that, we will illustrate how Northfield's approach facilitates the integration of unlisted assets in Total Portfolio Management and ERM. We will pay particular attention to examples of how liquid instruments can be used to manage the risk of illiquid assets, and will also demonstrate how illiquid asset classes can be taken out of their organizational silos and brought in a setting where the aggregate risk of all asset classes can be measured appropriately and accurately using a flexible ERM platform complemented by a broad factor risk model.
The Gorilla in the Room of Portfolio and Risk Management for Private Asset Classes: Business Cases for Institutional Real Estate and Infrastructure Portfolios
Presenters: Mihail Garchev, Guillaume Lavoie, and Alex Shannon, PSP Investments
May 14, 2014
Abstract
Risk factor portfolio management has become a prominent topic in the specialized institutional investment media. While many see risk factors as a superior diversification concept following the Global Financial Crisis, PSP Investments’ sees its fundamental value in the ability to develop a structured and disciplined investment process. In our view, this is particularly relevant for institutional investors with large allocations to alternative asset classes. The challenge for using risk factors lies in the ability of aggregating public and private investments in a consistent and coherent way and move away from the generalization and often times false comfort of proxies. Recognizing the deficiencies of commonly used approaches, we will be openly talking about the myths, fables, and inconvenient truths about portfolio and risk management for private asset classes. Our quest ultimately led us to a risk model which is deeply rooted in principles of valuation and links directly to the underlying investment, is able to capture various idiosyncrasies, and is flexible to incorporate a variety of assets (essentially any private asset) with simple or more complex structures.
This detailed bottom-up process allowed us not only to be able to understand and quantify the drivers of return for each individual asset, but also to assess the impact of macroeconomic developments, changes in real estate/infrastructure fundamentals; impact of acquiring or disposing assets; ability to hedge exposures; perform stress testing and scenario analysis; and assess the impact of various financing/leverage structures.
During the presentation, we will share with you a number of engaging short business cases from our Real Estate and Infrastructure portfolios that illustrate the main challenges and success stories of implementing the model for a diverse institutional real estate portfolio with a $20 bn of economic exposure.
Private Equity Real Estate Risk - The Good, The Bad, and The Ugly
Presenter: Rick Gold
May 13, 2014
Abstract
Whether you are an investor in a fund, directly own and operate properties, or some combination of the two, measuring the risk that these holdings add to your portfolio presents some unique challenges. This is becoming especially important as the number of large long-term institutional investors increase their allocations to illiquid assets such as real estate, private equity and infrastructure. Unfortunately, traditional appraisal-based risk and return metrics are fraught with statistical shortcomings that compromise both intra- and inter-asset class risk metrics. Over the past ten years, Northfield has been promoting an innovative property-level approach to measuring and managing directly owned real estate risk using a widely accepted factor based risk driver framework. Unlike other methodologies, Northfield's approach operates without the use of appraisal-based indices but remains firmly based in both real estate and modern portfolio theory fundamentals. This presentation will review Northfield's unique approach, and provide specific examples using a small real estate portfolio. It will also address the problems of traditional appraisal-based indices and why "pure-play" and REIT proxy approaches fall short.
Factor Representation of Asset Allocation and Portfolio Risk Inclusive of Illiquid Investments
Presenter: Dan diBartolomeo
May 13, 2014
Abstract
There have been two important trends in recent years with respect to the asset allocation practices of many large long-term investors such as sovereign wealth funds, pension schemes and university endowments. The first trend is that many have shifted their asset allocations to include very large commitments to illiquid investments such as real estate and private equity. Secondly, some asset owners are now thinking of asset allocation in terms of factor exposures (e.g. inflation) that transcend traditional asset class definitions. This trend appears to be a response to the extreme influence of macroeconomic factors during the financial crises of the past few years, and to the extremely large size of some funds which makes tactical shifts in asset allocation more difficult. This presentation will frame the fact that when investors invest in illiquid assets they are now dealing with new forms of risk. The more obvious problem is that the asset owner may be unable to liquidate illiquid assets into cash to meet consumption expenditures. A more subtle but economically more important problem is that the investor is losing the "option to rebalance" the portfolio from time to time leading to sub-optimal allocations. In addition, these asset owners formally ignore the volatility and correlation of such asset classes as either irrelevant or unable to be reliably estimated, leading to sub-optimal asset allocations with arbitrary default weights (e.g. 5% for real estate and 5% for private equity). In almost all cases, the underlying economic drivers between illiquid and traded asset classes are equally left unexplored.
PRMIA Webinar: Introducing Illiquid Investments to Total Portfolio Management and ERM
Presenter: Emilian Belev
April 29, 2014
Abstract
The rigorous risk analysis of illiquid investments like real estate and infrastructure alongside publicly traded investments in an ERM setting has been an object of many attempts, with little practical success until now. This presentation will outline the details of a new risk model designed for this task. The model is based in the philosophy of breaking complex investments down to their elemental pieces, and intuitively reassembling them into a coherent whole, using a general risk factor framework. The immediate benefits of this approach are two: a) it enables the rigorous risk measurement and estimation of the relevant factor exposures of investments with unobservable price time series, and b) it allows the seamless integration of measured risks of liquid and illiquid investments. We present evidence supporting the model results, and share the actual experience of a large scale implementation at a major pension fund.
Generating Tax "Alpha" for Private Wealth Households.
Presenter: Dan diBartolomeo
April 10, 2014
Abstract
It has been well established in the investment literature that it is possible to strategically control the impact of capital gain taxes, and thereby generate an after-tax "alpha" compared to the same investment strategy operated without regard to taxation. In this presentation we will review the multiple aspects of portfolio construction needed to create economically material tax alpha. Among the topics will be both capital loss and capital gain "harvesting" strategies, and the handling of dividend and mutual fund pass through taxation. We will also consider the impact of taxation on investor risk aversion, and making appropriate trade-offs between tax deferral and the expectation of active management alpha through explicit valuation of the "tax timing option" across different asset classes. Finally, we will illustrate a tax optimization broken into stages so as to make the process less of a "black box" to traditional investment practitioners.
Portfolio Management For Private Taxable Wealth: Basic Concepts and Operations
Presenter: Dan diBartolomeo
April 9, 2014
Abstract
Since our breakthrough release of tax-sensitive optimization methods nearly twenty years ago, Northfield has continued to research ways to improve both the investment outcomes for private wealth investors, and the operating profitability of asset managers that serve private clients. Much of this research was summarized in a 2006 CFA Research Foundation monograph (diBartolomeo, Horvitz and Wilcox). In this presentation, we will compare and contrast the recommended methods with less sophisticated "rule of thumb" approaches that have been adopted by some financial service organizations. We assert that such short-cuts in the characterization of investor objectives, the risk analysis of securities and funds, and the incorporation of taxes into portfolio construction dilutes the benefit of separate account management to the point of being a marketing concept without a material degree of investor benefit. The presentation will conclude with a discussion of the often abused concept of "house-holding," wherein the investment objective is to provide an optimal portfolio for a family or other set of related parties.
Non-parametric Methods for Asset Allocation in Private Wealth
Presenter: Dan diBartolomeo
April 8, 2014
Abstract
The Analytic Hierarchy Process is a powerful non-parametric method for making optimal decisions in a wide array of problems including the asset allocation. It is particularly well suited to the complex objectives of high net-worth households. Asset allocation portfolios obtained from the AHP approach have a high degree of robustness which makes them an ideal "prior" in Bayesian techniques intended to address solution instability in mean-variance optimized portfolios. To the extent that AHP brings focus to the needs and wants of the investor, rather than on our expectations of capital markets, the technique can be usefully employed to help quantify investor risk aversion which is otherwise unobservable. Finally, we can combine our improved specification of investor risk aversion to bring clarity to the multi-generational endowment aspects of the financial situations of high net-worth households.
There’s More to Evaluating Risk in Real Estate Portfolios than Location, Location, Location!
Presenter: Rick Gold
March 5, 2014
Abstract
Whether you own a REIT that focuses on hotel properties or a building in the center of Hong Kong, measuring the risk that these holdings add to your portfolio presents some unique challenges. Northfield has created risk models and processes to assess the volatility of these holdings. This webinar will provide an introduction to the differences between public and private real estate investments using both debt and equity instruments. Rick will then delve into the factors that are used to evaluate the risk in REITs in the US and Globally and discuss how those factors were derived. Finally, Rick will show how private real estate holdings are being evaluated within a multi- asset class framework.
This webinar is suitable for anyone who uses real estate as part of their investment portfolio and is particularly appropriate for those managing either public or private real estate portfolios.
Understanding Risk Decomposition
Presenter: Dan diBartolomeo
January 29, 2014
Abstract
Risk models help professional investors quantify how much risk they are taking in their portfolios on either an absolute or relative basis. This webinar is designed to help both quants and fundamental investors understand the results that they see in their risk decomposition reports. Dan will first review the basic structure of the risk model and how that will influence the breakdown of risk. He will look at how the results will vary based upon the number and type of factors in the model. From there, he will discuss the alternative ways that various vendors break down risk. Finally, he will look at the perils of over specifying the sources of risk to the position level.
Attendees will go away with the ability to distinguish which sources of risk in their portfolio are acceptable and which should be a concern. They will also be able to recognize the differences between alternative vendor reports and their effect on the results.
Portfolio Construction – Optimizer Advantages, Hazards and Tips
Presenter: Dan diBartolomeo
October 22, 2013
Abstract:
The more powerful the tool, the more dangerous it becomes when used improperly. This is true for lawn mowers and chain saws, and is definitely true for portfolio optimizers. Most tools will allow the users to do the job better in significantly less time. Using an optimizer, portfolios can better balance the expectations of the portfolio manager providing they are properly employed. With over 30 years of working with clients, Dan diBartolomeo has worked with hundreds of portfolio managers to create many different types of portfolios using an optimizer. To say the least, he has learned a few tricks along the way. This webinar is aimed at providing the attendee a sound understanding of how an optimizer works and what it is capable of doing and not doing. It is aimed at both the novice and somewhat seasoned investor who wants to get more out of their current investment process. In addition to reviewing the basics, Dan will also discuss some interesting challenges that he has faced in the past and show how these were resolved. He will also share his top ten things not to do when optimizing a portfolio.
New Features and Functions of Northfield’s Performance Attribution Service
Presenter: Tracy Licklider
March 20, 2013
Abstract:
A major new release of Northfield’s Performance Attribution service will be made available in the next few weeks. The key difference between the new version and the predecessor is an extensive new capability to produce presentation quality reports, graphs and charts using an embedded version of the Crystal Reports report generator. The new version takes the very extensive reporting output of our analysis and automatically loads it into commonly used SQL databases including systems like MySQL which are available to end users at little or no cost. The Crystal Reports system can then be used to retrieve the data from the database and generate an essentially infinite variety of user defined reports and graphics.
The Performance Attribution system has also had a variety of "behind the scenes" enhancements that allow it to handle analysis of march larger problems as might be common in the context of back-testing for investment research. For example, capacity is now sufficient to handle a twenty year monthly back-test involving security universes with a thousand or more members. Later in 2013, a new GUI consistent with our desktop Open Optimizer will also be added to this application.
This webinar will discuss the new features and offer an in-depth demonstration of the new Crystal Reports Report Generator.
Liquidity Planning Tools and Strategy Capacity for Equity Markets
Presenter: Dan diBartolomeo
January 23, 2013
Abstract:
Liquidity shortages during the Global Financial Crisis underlined the need for both asset managers and institutional investors to have well articulated liquidity policies as part of their strategic investment planning. The first aspect of such policies should be to understand the potential liquidity needs of a fund and the cost of carrying out a partial liquidation under difficult market conditions. We will present a simple metric for “liquidity-adjusted risk” to complete this first aspect of our discussion. The second part of the presentation will provide a model of AUM capacity of particular active strategies in terms of the tradeoffs between turnover, trading costs and alpha. We will review model output in light of the empirical data provided in Elton, Gruber and Blake (2011). Both of these procedures require framing the cost of transacting equity trades into true supply/demand curves that have quantity, side and time dimensions. The presentation will conclude with a demonstration of a new “liquidity planning tool” software application will shortly be available to Northfield clients free of charge.
Third Generation Northfield Risk Models
Presenter: Anish Shah
November 8, 2012
Abstract:
While Northfield continually improves and develops new products, e.g. the Adaptive Near Horizon models, eight years have passed since the last major update of the existing models. With testing nearing completion, the next generation of Northfield models will soon be released to clients. The first part of the presentation describes the considerations that led to updating the models. Next is a technical overview of the econometric and structural changes. Last are samples of second and third generation model data to illustrate the benefits of the update.
Wealth Management, Investor Suitability, Fiduciary Requirements and Financial Regulation
Presenter: Dan diBartolomeo
September 19, 2012
Abstract:
In recent weeks, rules approved by the US SEC in 2010 went into effect mandating much higher standards for financial services firms in ensuring that investor suitability guidelines are followed in dealing with individual investors. While additional training of staff dealing with retail investors can help, we assert that substantial conformity with the new guidelines can only be achieved through investment technology. To guarantee and document compliance with the regulations, such systems must bring sophisticated analysis to bear on the problem of giving sound financial advice to investor households, while being simple and inexpensive to implement in the field.
This presentation will focus on three analytical methods that Northfield has pioneered in our wealth management practice. The first is the Analytical Hierarchy Process, a questionnaire-based industrial management technique that can provide very robust decisions on investor asset allocation and financial product selection, as described in Bolster and Warrick (Journal of Wealth Management, 2008). The second is the implementation of the Discretionary Wealth Hypothesis (Wilcox, 2003) which provides an analytical solution to how investor risk tolerance should vary across time and across households with differing degrees of financial strength. Finally, we will discuss a more complete asset/liability management structure from diBartolomeo (2011) that incorporates an arbitrage-free term structure model to align asset allocation choices to the type and timing of household contingent liabilities for future consumption expenditures.
How to use the Northfield Optimizer in R & MATLAB®
Presenter: Peter Horn
April 3, 2012
Abstract:
At Northfield we are committed to enhancing the accessibility and usability of our analytics and to support these goals we are investing in a new service called the Northfield Analytics Services Platform. The first phase has been completed and enables clients to access the Northfield Risk Models and Optimizer Service from the R, MATLAB & Java programming environments. This workshop will demonstrate how you can use the Northfield Optimizer in your preferred programming environment.
Recent Product Enhancements from Northfield Research
Presenter: Anish Shah
September 13, 2011
Abstract:
Northfield will soon release the next generation of risk models. What's new and why was it done? This webinar describes the issues motivating both the models and the most recent enhancements to the Northfield Open Optimizer.
A Detailed Examination of Minimum Variance and Low Volatility Equity Strategies: A Real Market Inefficiency or Sleight of Hand
Presenter: Dan diBartolomeo
July 12, 2011
Abstract:
Despite a rich academic literature that goes back to the early 1990s, equity portfolio strategies based on selecting low volatility securities or forming minimum variance portfolios have become pervasive in recent years. This webinar will begin with an examination of various theoretical reasons why such strategies may be very attractive for many investors, but not actually represent real market inefficiencies. These reasons include commonly used but incorrect specifications of the CAPM, and the often overlooked distinction between arithmetic mean and geometric mean of observed returns. We will then present a ten year empirical analysis of minimum variance strategies across US, European and Asian equity markets. The final aspect of the presentation will be to conduct comparable examination of equal weighted, capitalization weighted, and minimum variance portfolios selected from securities ranked either high (safe) or low (risky) on Northfield’s corporate sustainability mode (published Journal of Investing December 2010), so as to allow us to distinguish between the impact of security selection and portfolio formation.
Optimization for Northfield Users
Presenter: Mike Knezevich
May 10, 2011
Abstract:
This online workshop was a Northfield-centric description of optimization. We built upon the tenets of modern portfolio theory as it is incorporated in Northfield’s Optimizer. The maximization of the utility function was discussed, subject to different constraints and how the Optimizer mitigates constraint conflict during the optimization process. Although the content was theoretical in nature, this Workshop was best suited for those familiar with the Northfield Optimizer.
Key Elements of Risk Control for Asset Managers
Presenter: Richard Pearce
March 8, 2011
Abstract:
This Webinar Covers:
-Multiple reasons for using a risk model
-How risk models are built: model types, factor choices and completeness, time horizon
-Northfield risk models: analysis of reports and their part in portfolio construction
Lies and Performance Attribution
Presenter: Steve Gaudette
January 25, 2011
Abstract:
Performance attribution can be used in many ways to hide the truth behind what is going on in a portfolio by cherry picking the data that is produced. This workshop will review several approaches to uncover the truth behind the attribution.
Redefining Private Equity Real Estate Risk
Presenters: Emilian Belev and Rick Gold
December 7, 2010
Abstract:
Private equity real estate has been hampered by its dependency on appraisal-based valuations and the lack of transaction-based pricing. Appraisal bias not only dampens return volatility but also produces well known distortions in portfolio optimization allocations. The results are portfolio allocations that are artificially and inefficiently determined. In addition, to its REIT factor models; Northfield has also developed a private equity risk model which circumvents the need for appraisal-based valuations by directly forecasting return variance at the property level rather than the traditional approach of forecasting the level of expected returns. By estimating variance at the property and ultimately portfolio levels, user-provided estimates of expected returns can be provided to form an efficient frontier either within real estate or across asset classes. The model can be used to determine whether buying an office building in San Jose or Seattle enhances diversification given the intersection of their tenant and stock profiles. It can also help them understand the implications of fixed or variable rate financing given the structure of their prepayment options in their bond portfolio. This talk will review the model’s basic framework and discuss some of the model’s other interesting ancillary benefits.
The Central Paradox of Active Management
Presenter: Dan diBartolomeo
October 19, 2010
Abstract:
Within asset management, the risk of benchmark relative performance is typically expressed by measures such as “tracking error”, which describes the expectation of times-series standard deviation of benchmark relative returns. This is clearly a useful measure for index fund management, where the expectation of the mean for benchmark relative return is fixed at zero. The active management case is problematic, as tracking error excludes the potential for the realized future mean of active returns to be other than the expected value. All active managers must believe their future returns will be above benchmark (or peer group average) in order to rationally pursue active management, yet it is axiomatically true that roughly half of active managers must produce below average results. Following the convention of Qian and Hua (2004), we refer to this additional portfolio risk as “strategy risk”. In this presentation, we will first provide both theoretical and empirical approaches to estimating the magnitude of strategy risk across asset classes and manager styles. We will then illustrate procedures to incorporate this additional risk into security level portfolio optimizations, risk budgeting and manager selection procedures.