Functional False Discovery Rate in Mutual Fund Performance

Abstract

We introduce a novel multiple hypothesis testing framework for selecting outperforming mutual funds with control of luck, the functional False Discovery Rate “plus”. We show that our method, which incorporates informative covariates to estimate the false discovery rate, gains considerable power over the False Discovery Rate “plus” of Barras, Scaillet and Wermers. We experiment with five covariates that commonly affect the mutual funds’ performance by constructing portfolios that generate positive alphas. Our results confirm the informative power of the five covariates, and we demonstrate their economic value in mutual funds’ selection after controlling of luck.