Empirical Characteristics of Dynamic Trading Strategies: The Case of Hedge Funds

Motivation(s)

Most mutual fund managers have investment mandates similar to traditional asset managers with relative return targets. They are typically constrained to hold assets in a well-defined number of asset classes and are frequently limited to little or no leverage (quantity). Their mandates are to meet or exceed the returns on their asset classes. These restrictions enabled Sharpe to model the stylistic and performance differences between managers.

Another class of managers are hedge fund managers and commodity trading advisors (CTAs). Their mandates are to make an absolute return target regardless of market environment. In order to meet the mandates, they are allowed to choose among many asset classes and utilize dynamic trading strategies (e.g. short sales, leverage, derivatives). Unfortunately, this newfound flexibility is not accounted for in Sharpe’s asset class factor model.

The only style analysis publicly disclosed by the hedge fund industry is a categorization of trading strategies:

Systems/Trend Following

Describes managers who uses technical trading rules.

Systems/Opportunistic

Technically driven traders who also take occasional bets relying on rule-based models.

Global/Macro

Managers who primarily trade in the most liquid markets in the world (e.g. currencies, government bonds) typically betting on macroeconomic events (e.g. changes in interest rate policies, currency devaluations) and relying mostly on their assessments of economic fundamentals.

Value

Traders who buy securities of companies they perceive to be undervalued based on their microanalysis of the fundamentals.

Distressed

Managers who invest in companies near, in, or recently emerged from bankruptcy/corporate restructuring.

Proposed Solution(s)

The authors propose adding new factors to Sharpe’s model to capture the absolute return investment style.

Running Sharpe’s style regression on the returns of hedge funds revealed low and sometimes negative correlation with asset class returns, which is reflective of how leverage (quantity) can influence what the regression coefficient converges to.

The additional style factors provide an analytical framework for managing portfolios with a better diversity of styles. Unlike the hedge fund industry’s qualitative descriptors, factor analysis can quantify the actual returns of these investment styles. The empirical results show that style diversification can improve the performance of a traditional stock-bond portfolio without substantially increasing its risk.

Evaluation(s)

The authors analyzed a newly compiled dataset consisting of 3,327 US mutual funds and 409 hedge funds/CTA pools. Each fund had at least 3 years of monthly returns and the non-mutual funds were required to have at least $5 million in assets under management.

Applying the proposed model on the mutual funds reaffirmed the high level of correlation between mutual fund returns and asset classes, which indicates that mutual fund styles are basically buy-and-hold strategies utilizing various asset classes.

The analysis of the hedge funds requires returns standardization (i.e. scale to zero mean and unit variance) to remove differences in variances caused by leverage differences (e.g. two funds employing the exact same trading strategy but different leverage will have different return variances).

Treating the hedge funds as a single group, the authors extracted five mutally orthogonal principal components, which approximately explained 43% of the cross-sectional return variance. For each of the style factors, the author formed a portfolio using hedge funds/CTA pools that are correlated only to that principal component. The portfolio weights were chosen to maximize the correlation between the portfolio returns and the corresponding principal component.

Since dynamic trading strategies deal with extreme or tail events, the authors focused on economical significance instead of statistical significance. They divided the monthly returns of each asset class into five environments ranging from severe declines to sharp rallies. The empirical evidence shows that only the Value style is akin to buy-and-hold. The Distressed style is simply sensitive to high yield corporate bonds and is not a buy-and-hold. The other styles are not sensitive to the asset markets in the normal states, but can be sensitive to selective markets during extreme states.

The authors warned that the proposed styles do not cover arbitrage strategies (e.g. short sellers, spread traders) and should not be blindly used for to determine performance attribution.

Future Direction(s)

  • Since hedge fund managers are not required to disclose their performance and assets under management publicly, can data mining uncover this information?

  • Are market neutral strategies more aligned with volatility trading or index funds?

  • How to predict diversification implosion with existing data mining techniques?

  • How much does survivorship bias influence the returns of the proposed styles?

Question(s)

  • Not completely sure how the style factors were computed.

Analysis

Incorporating style factors is one way to quantify the actual returns of hedge funds. The methodology is still debatable: at the end of the day, does it not boil down to risk, money, and returns? It is not obvious why one needs to establish different categories. While the empirical results were reasonable, the paper would have been more convincing if there was some comparison against momentum strategies. It is quite interesting that performance attribution requires such deep thoughts. One would think if an algorithm can achieve similar gains, then the net difference is the manager’s skills? Fortunately, hedge fund managers need to recover all losses before getting their commission. Even though the authors did not freely make available their dataset, one of the author’s homepage has other datasets that could be useful for backtesting.

References

FH97

William Fung and David A Hsieh. Empirical characteristics of dynamic trading strategies: the case of hedge funds. Review of financial studies, 10(2):275–302, 1997.