Finance Department


Date of this Version



Quantitative Finance, 2015 Vol. 15, No. 10, 1705–1726,


US govt


This study tests the performance of 14 hedge fund index clones created using parsimonious outof- sample replication portfolios consisting solely of easily accessible assets. We employ a genetic algorithm to integrate two traditional hedge fund replication methods, the factor-based and payoff distribution replication methods, and evaluate over 4500 commonly held stocks, bonds and mutual funds as replicating portfolio components. In-sample performance indicates that hedge funds have return series similar to portfolios of commonly held assets, and out-of-sample results provide evidence that the in-sample relationships can hold with infrequent rebalancing. This hedge fund replication attempt rates well relatively to prior efforts as 11 replicating portfolios have out-of-sample correlation values of at least 60%. Overall, these results show promise for using a genetic algorithm technique to replicate hedge fund returns.