Tracking Hedge Fund Performance: A Balanced Sampling Strategy

This paper discusses the application of statistical, survey sampling technique to hedge fund tracking problems. I describe a strategy that allows an investor or a fund of hedge funds manager, to construct a small tracking portfolio that replicates the time series changes of the total relative Net-Asset-Value (NAV) of a large basket of funds. The trackers are constructed using a method of balanced sampling, in which components are selected randomly with unequal inclusion probabilities. Empirical studies are performed on directional hedge fund styles: Commodities trading advisors, Global macro and Equity hedge. In all cases, empirical results show that the proposed strategy replicated efficiently the total fund's relative NAVs using only ten percent of the sample. The constructed portfolios are stable in the long run, allowing the investor to implement a passive investment strategy in the alternative investment universe. I also consider a larger sample of funds, mixing the aforementioned category of hedge funds. The market tracking ability of balanced sampling remains statistically significant.