Skilled Mutual Fund Selection: False Discovery Control under Dependence

Selecting skilled mutual funds through the multiple testing framework has received increasing attention from finance researchers and statisticians. The intercept $\alpha$ of Carhart four-factor model is commonly used to measure true performance funds, positive $\alpha$'s are considered as skilled. We observe that standardized OLS estimates across possess strong dependence nonnormality structures, indicating conventional methods inadequate for selecting funds. start a decision theoretic perspective, propose an optimal procedure minimize combination false discovery rate non-discovery rate. Our proposed constructed based on probability each fund not being conditional information all in our study. To distribution procedure, we consider mixture under new method called "approximate empirical Bayes" fit parameters. Empirical studies show selected have superior long-term short-term performance, e.g., selection strongly outperforms S\&P 500 index during same period.