Hedge Funds
Member Library
This paper introduces a number of quantitative tools for manager selection, due diligence, and the ongoing monitoring of hedge funds that the authors believe to be particularly suited to the nature of hedge fund returns. The analyses introduced typically address information relevant to the entire distribution function, as compared to the more conventional practice of utilizing lower moments such as mean returns and standard deviations. The authors start by illustrating the degree to which dispersion occurs in hedge fund returns, indicating the need for a high level of both quantitative and qualitative manager selection. They note a strong relationship between volatility and the inter-quartile return statistic that may be a result of the leveraged nature of the strategies involved. They then test the hypothesis that hedge fund performance will persist, suggesting through parametric and non-parametric tests that it is flawed. However, they find by comparison that risk, as defined by volatility, is highly persistent and conclude that this strongly supports a risk budget approach to hedge fund allocation, offering a new, objective algorithm for risk budgeting. For manager selection, they propose the use of the Hurst exponent in conjunction with the D-statistic to identify managers with persistent good performance, also introducing a formal methodology, utilizing a logit model to isolate the characteristics of a database of liquidated funds and find agreement with an earlier model proposed for liquidation. Finally, noting that while a manager's returns, standard deviation, downside deviation, and a number of other statistics should be analyzed, they find the & function has particular relevance to hedge funds given its statistical equivalence to the return series and its sensitivity to the investor utility function.