Outlining an equal volatility-adjusted approach to hedge fund management
The hedge fund industry has come to represent a significant portion of institutional portfolios, with about $3.2trn (€2.8trn) in assets under management (AUM) in the second quarter of 2019. Academic and industry publications have provided valuable insights into certain aspects of portfolio management in this field, particularly in the area of fund evaluation and portfolio construction. Nevertheless, a robust and flexible methodology capable of evaluating whether those insights can benefit a specific institutional investor that is subject to real-world constraints is lacking.
There are several important challenges that require careful consideration when trying to assess benefits of adding hedge-fund investments to a traditional portfolio:
● First, investors have objectives that vary substantially, depending on the type of institution they represent. For example, an asset management firm might seek to maximise the Sharpe ratio. In contrast, a university endowment attempts to achieve returns that exceed the university’s spending rate over a market cycle, while a pension fund tries to maximise risk-adjusted return within an asset-liability framework.
● Second, sophisticated investors often utilise rigorous filtering criteria such as length of track record or level of assets under management. Most academic studies either ignore these selection criteria or selectively incorporate them with the purpose of accounting for certain biases such as the small-fund bias or the incubation bias. While accounting for biases is important, an institutional investor ultimately wants to assess the benefits of making changes to his portfolio that satisfy his own set of preferences and constraints.
● Third, most academic papers often compare portfolios that include hundreds of funds which might be irrelevant to an investor who plans to allocate to three to five hedge funds. Such an investor would be interested in evaluating the impact of his manager selection and portfolio construction decisions on the distribution of outcomes. Generating out-of-sample results for randomly selected subsets of five manager portfolios within a simulation framework would provide this information.
● Fourth, while the investor cares about the marginal impact of the hedge fund investment on his existing portfolio, this impact is often ignored in traditional analyses.
● Finally, hedge fund databases provide returns that are delayed by about one month. This delay is usually ignored in academic papers, and creates a significant barrier to implementing the results of most studies.
A simulation-based methodology for evaluating hedge fund investments
In our research paper, ‘A simulation-based methodology for evaluating hedge fund investments’, published in the Journal of Asset Management (2016) a methodology designed to evaluate hedge-fund investments is introduced, subject to the realistic constraints that institutional investors face. The methodology can be customised to the real-life preferences and constraints of investors. These include investment objectives, performance benchmarks, desired number of funds in a portfolio and rebalancing frequency.
This methodology can be illustrated by imposing the framework on a dataset of commodity trading advisors (CTAs) with 604 active and 1,323 defunct funds over the period 1993-2014. The out-of-sample performance of three hypothetical risk-parity portfolios and two hypothetical minimum-risk portfolios and their marginal contributions to a typical 60-40 portfolio of stocks and bonds can then be measured.
“An institutional investor ultimately wants to assess the benefits of making changes to his portfolio that satisfy his own set of preferences and constraints”
The conclusions are:
● Equal-risk portfolio construction methodologies dominate minimum risk ones. Minimum-risk portfolios perform the worst across all performance metrics considered (absolute performance, Sharpe ratio, Calmar ratio). Their average Sharpe ratios are between 0.299 and 0.304, lower than the 0.319 average Sharpe ratio of the random portfolio and the average Sharpe ratios of 0.342 to 0.362 of the equal risk methodologies.
● An investment in CTAs improves the performance of the 60-40 portfolio regardless of the choice of portfolio construction approach. For the out-of-sample period between January 1999 and December 2014, a 10% allocation to managed futures improves the Sharpe ratio of the original 60-40 portfolio of stocks and bonds from 0.376 to 0.399-0.416 on average, depending on the portfolio construction methodology. Blended portfolios have higher Sharpe ratios in at least 89% of simulations and higher Calmar ratios in at least 89.5% of simulations.
The question then arises as to how much a 60-40 portfolio should be allocated to CTA managers to make it optimal? The figure shows that the average Sharpe ratio of a blended portfolio reaches its highest value of 0.507 at 40% allocation to equally-weighted portfolios of five CTAs. This is a substantially higher value than the 0.376 Sharpe ratio of the original 60-40 portfolio of stocks and bonds.
The average Calmar ratio of a blended portfolio reaches its highest value of 0.24 at 60% allocation to equally-weighted portfolios of five CTAs. This value is substantially higher than the 0.092 Calmar ratio of the original 60-40 portfolio of stocks and bonds.
Portfolio management with drawdown-based measures
Finally, starting from the observation that an investor cares more about her portfolio’s drawdowns than its volatility, we explore in a second paper, ‘Portfolio management with drawdown-based measures’, published in the Journal of Alternative Investments (2017), a newly-introduced drawdown risk metric, the modified conditional expected drawdown (MCED).
Sophisticated investors often track a popular maximum drawdown (MDD) measure that shows the maximum peak-to-valley loss of an investment because it represents the worst-case scenario for an investor who invests at the top and redeems at the bottom. However, historical MDD is based purely on actual performance and gives no indication of what could have happened in a slightly different environment.
The proposed MECD metric is designed with the goal of measuring a portfolio drawdown potential and a contribution of each portfolio component to the drawdown. Our equal-risk contribution MCED approach resembles a popular risk-parity approach by constructing a portfolio with each portfolio constituent contributing equally to the MCED of the portfolio. The new portfolio technique is assessed using the previously introduced evaluation framework. This analysis of Sharpe and Calmar ratios shows that while it dominates all other drawdown-based techniques considered, it still fails to consistently outperform the simple equal-volatility-weighted approach.
In our research papers, a quantitative large-scale simulation framework for robust and reliable evaluation of hedge fund investments with real life constraints faced by institutional investors is introduced and illustrated. This methodology is implementable and incorporates common investment constraints when creating and rebalancing portfolios. The framework is customisable to the preferences and constraints of individual investors, investment objectives, rebalancing periods and the desired number of funds in a portfolio and can incorporate many portfolio construction approaches. Thus, the methodology can benefit portfolio managers, investment officers, board members and consultants who make hedge fund investment decisions. Using this framework, different portfolio construction methodologies can be compared, including one based on a newly-designed maximum drawdown measure, the modified conditional expected drawdown. Our results show that the equal volatility-adjusted approach delivers the best risk-adjusted performance.
Marat Molyboga is the head of research and chief risk officer at Efficient Capital Management. Christophe L’Ahelec is a senior principal in the external managers group, capital markets, at the Ontario Teachers’ Pension Plan