Evaluating mutual fund performance is a topic of long-standing interest in the academic literature, but few if any studies have addressed the selection of an optimal portfolio of funds. Instead of using the historical data to estimate performance measures or produce fund rank- ings, this study uses the data to explore the mutual-fund investment decision. Speci¯cally, from an investment universe of over 500 no-load equity funds, we construct portfolios having the ex ante maximum Sharpe ratio based on a Bayesian predictive distribution that combines the information in historical returns with an investor's prior beliefs, accounting for param- eter uncertainty. .
