Research Spotlight | Using AI to Identify Skilled Managers

December 11, 2023

The Bottom Line: A new study confirms that investors can identify skilled managers in advance and that they reallocate their investment dollars towards those skilled managers.

The Study: “Machine-Learning the Skill of Mutual Fund Managers” by Ron Kaniel, Zihan Lin, Markus Pelger, and Stijn Van Nieuwerburgh. NBER Working Paper 29723, published July 2022.

The Process: The authors use a neural network to predict mutual fund outperformance, by examining actively-managed U.S. equity funds over the 40-year period from 1980 to 2019. They seek to predict fund returns using a large set of factors, including both characteristics of the fund itself and characteristics of the stocks held by the fund. By incorporating measures of investor sentiment and of macroeconomic activity, they also study of the role of “big picture” factors in mutual fund returns.

The authors divide the data into three periods of equal length using two different methodologies. Then using two of the three periods to develop the model, they make predictions for the third period for each of the methodologies.

The Findings: The model is able to identify skilled funds and predict outperformance. Investing according to the model’s predictions generates a cumulative outperformance (above Treasury bill returns) of 72% for the top decile over the period. Avoiding the bottom decile is an even more profitable strategy given that these funds generated an underperformance of 119% over the same period.

The outperformance is longer-term, with the best decile outperforming the worst decile for three years.

Of all the fund characteristics studied, two have a particularly significant role in accurate return prediction: fund momentum (meaning performance over the previous year) and fund flow (meaning the change in the fund’s assets in the past month). These factors are especially predictive when investor sentiment is high.

Incorporating a measure of macroeconomic activity increases the accuracy of the model overall.

As the authors summarize, “Skill, therefore, leaves a trail in the form of fund return momentum, and investors can exploit this to earn higher returns.”

The Implication: This study helps to explain why asset-weighted fund returns have been generally higher than average fund returns; namely, investors can identify superior performance in advance and reallocate their assets into funds with that superior performance, as predicted by Berk and Green in their 2004 paper, “Mutual Fund Flows and Their Performance in Rational Markets.” The results are consistent with the findings of previous studies of manager return prediction, which are summarized by Cremers, Riley, and Fulkerson in 2019 in “Challenging the Conventional Wisdom on Active Management: A Review of the Past 20 Years of Academic Literature on Actively Managed Mutual Funds.”

The study also highlights the role of fund marketing. The authors note that marketing-driven flows can combine with flows driven by performance to create a “virtuous circle” for funds.