Seminar (Dept. of Finance)

Title: Selecting Mutual Funds From the Stocks They Hold: a Machine Learning Approach

Speaker: Bin Li, Professor, Wuhan University

Time: 10:00-11:30, June 20th , 2025 (Friday)

Venue: Room 1007, Mingde Business Building

Language: English/Chinese

ABSTRACT:

We combine individual mutual fund holdings and a large number of stock characteristics to compute fund-level exposures to stock characteristics and show that machine learning methods, particularly boosted regression trees (BRTs), outperform traditional models in predicting mutual fund performance by capturing non-linear relationships, interactions, and factor momentum. Our findings reveal that mutual fund managers’ performance can be captured by their ability to tilt portfolios toward stock characteristics that deliver superior returns, with a sizable fraction persistently outperforming others in this dimension. These results highlight the importance of dynamic characteristic exposures in explaining mutual fund performance over traditional static benchmarks.