Evaluating Hedge Fund Performance

Abstract

The search for methodologies that accurately measure performance and performance persistence continues to evolve. This is especially true for investment strategies such as hedge funds, which have been shown, in several instances, to not be normally distributed. In this article, we evaluate performance of hedge funds using conditional approaches and GMM. Unlike the Sharpe ratio or Jensen's alpha, our results would still be valid even if hedge funds were not normally distributed. We use the CISDM hedge fund database for this study. We create three portfolios to measure performance: an Active portfolio (which consists of funds in the active database), a Dead portfolio (which consists of funds in the defunct database), and an All portfolio (which consists of funds in both the active and defunct databases). We find that while the Active portfolios show evidence of positive risk-adjusted returns in most cases, the Dead portfolios do not and only some of the All portfolios show evidence of positive risk-adjusted returns. The results are similar irrespective of whether we use Jensen's alpha or conditional approaches. Our results point to two conclusions: one, the explanatory variables used in this study may not be able to capture the type of trading strategies followed by hedge fund strategies and, two, the estimated alphas are good estimates of the true alphas which are mostly due to managers' skills and hence cannot be explained by naïve static or dynamic trading strategies. In our analysis of market timing models, we show that hedge fund managers in general lack market timing ability and fund level analysis is required to determine the few that do have market timing ability. The results also suggest that hedge fund returns have option-like properties and future research should include option-based factors in performance evaluation.