Volume 30 Number 4, 2025

Informing Private Equity Selection for Limited Partners using Machine Learning

Daniel R. Cavagnaro ^a, Yinfei Kong ^b, Yingdi Wang ^c, ^*
a,b Department of Information Systems and Decision Sciences, California State University, Fullerton 
c Department of Finance, California State University, Fullerton 
a dcavagnaro@fullerton.edu 
b yikong@fullerton.edu 
c yingdiwang@fullerton.edu
 

ABSTRACT

We evaluate whether limited partners can use machine learning on structured data to help identify top-performing private equity funds. We train and test six supervised learning algorithms (linear probability, regularized and additive, and tree-based models) on a sample of 1,402 venture and buyout funds raised between 1990 and 2011. We find that using these models to select investments can generate economically large returns. While the models vary in their ability to identify top performers accurately, they generally outperform benchmark heuristics in out-of-sample tests. The strongest gains are found for buyout funds and for the combined sample. Results for venture funds show less pronounced benefits. We further find that machine learning considers multi-dimensional interaction terms, rather than single fund or firm characteristics, important for predicting returns. Overall, our results suggest that machine learning can serve as a tool to aid limited partners in their decision-making process, particularly in buyout fund selection.

JEL Classifications: G11, G23, G24

Keywords: private equity, machine learning, institutional investors

Cite this article: 

Cavagnaro, D. R., Kong, Y., & Wang, Y., 2025, Informing Private Equity Selection for Limited Partners using Machine Learning, International Journal of Business, 30(4), 002. https://doi.org/10.55802/IJB.030(4).002