We present a new approach to selecting funds with superior risk-adjusted investment performance.
Our approach eliminates funds with unpredictable or inferior performance through a sequence of pair-wise comparisons that determines both the number of superior funds and their identity, while discarding funds with good past performance due to luck.
We apply our approach to a sample of U.S. domestic equity mutual funds using a conditional performance model that merges information from returns and holdings data and find that funds identified as being superior earn substantially higher risk-adjusted returns than top funds identified by conventional ranking (portfolio sorting) methods. Moreover, we find strong evidence of variation
in the breadth of the set of funds identified as superior as well as fluctuations in the style and industry exposures of such funds across time and economic states.

To read the paper click here.