Performance Persistence In Hedge Funds: A Bayesian Analysis Under Conditional Value-at-Risk Constraints∗

I study the performance of hedge funds portfolios and find persistence at three-year horizons. I show that, minimizing the Conditional Value-at-Risk at the portfolio formation leads to an annual increase in performance of 63 % to 120 % as compared to equally weighted portfolios. Moreover, I also find that using Bayesian alphas or t-statistics instead of OLS alphas as performance measures increases the predictability of the performance. My results are robust to incubation bias, backfilling bias, and are not serial correlation driven. Resulting from a CVaR minimization procedure, they are consistent with the risk constraints accruing to fund of hedge funds managers