北航经管学院“经济统计论坛”系列讲座
（2025年第6期，总第37期）
讲座题目：Can Mutual Funds Exploit the Anomaly Zoo? Evidence from a Customized Machine Learning Framework
讲座时间：2025年12月26日（周五），10:00-11:00
会议地址：新主楼 A618
讲座嘉宾：舒腾嘉 助理教授
邀请人：康雁飞 教授
讲座嘉宾
Tengjia Shu is an Assistant Professor of Finance in University of Illinois Chicago. She earned my Ph.D. degree in Business Administration (Finance) from University of Iowa. Her main research interests are investment, asset pricing, machine learning, and labor economics.
讲座概要
I propose a joint framework to study how mutual funds incorporate stock-level anomalies into portfolio choices. Rather than inferring skill solely from realized returns, our approach evaluates fund portfolios on a common anomaly payoff surface estimated from stock characteristics. I introduce an Anomaly Investing Measure (AIM) that summarizes the anomaly-implied return premium embedded in a fund’s active portfolio tilts, constructed using both realized and policy-implied holdings. Based on a flexible machine learning model, I jointly estimate portfolio policies and anomaly-related performance. I address the high dimensionality of characteristics with a selective principal component analysis without materially affecting predictive accuracy. Empirically, AIM exhibits substantial cross-sectional and time-series variation and predicts future out-of-sample excess returns and factor-adjusted alphas, even after controlling for alternative performance measures. Predictability is stronger in market stress periods.
