Efficient Fund of Hedge Funds Construction Under Downside Risk Measures

Abstract

We consider portfolio allocation in which the underlying investment instruments are hedge funds. Benchmarks and conditional-value-at-risk motivate a family of utility functions involving the probability of outperforming a benchmark and expected shortfall from another benchmark. Non-normal return vectors with prescribed marginal distributions and correlation structure are modeled and simulated using the normal-to-anything method. A Monte Carlo procedure is used to obtain, and establish the quality of, a solution to the associated portfolio optimization model. Computational results are presented on a problem in which we construct a fund of 13 CSFB/Tremont hedge-fund indices.

Keywords: Portfolio choice, expected regret, hedge funds, fund of funds, portfolio optimization, Monte Carlo simulation

JEL Classification: C15, C61, G11