Evaluating Hedge Funds with Machine Learning-Based Benchmarks

68 Pages Posted: 28 Sep 2022 Last revised: 23 Jan 2024

See all articles by Tengjia Shu

Tengjia Shu

University of Illinois at Chicago

Ashish Tiwari

University of Iowa

Date Written: September 9, 2022


The explanatory power of multi-factor models used to evaluate hedge fund performance is effectively zero for many funds (so-called zero-R2 funds). We explore alternative approaches to benchmarking hedge funds based on machine learning techniques. In general, machine learning algorithms allow for significantly improved performance tracking, especially for zero-R2 funds, resulting in more precise estimates of fund alphas and hence, more accurate identification of superior funds and fund failures. Our results offer compelling evidence that machine learning-based benchmark models can effectively capture the nonlinearities and interactions among risk factors, which are crucial for accurately characterizing risks associated with hedge fund strategies.

Keywords: Hedge Fund, Investment, Bayesian Inference, Machine Learning, Regression Trees

JEL Classification: G11, G23

Suggested Citation

Shu, Tengjia and Tiwari, Ashish, Evaluating Hedge Funds with Machine Learning-Based Benchmarks (September 9, 2022). Available at SSRN: https://ssrn.com/abstract=4215002 or http://dx.doi.org/10.2139/ssrn.4215002

Tengjia Shu (Contact Author)

University of Illinois at Chicago ( email )

601 S. Morgan St
Chicago, IL 60607
United States

Ashish Tiwari

University of Iowa ( email )

Finance Department
Henry B. Tippie College of Business, 108 PBB
Iowa City, IA 52242
United States
(319) 353-2185 (Phone)
(319) 335-3690 (Fax)

HOME PAGE: https://tippie.uiowa.edu/people/ashish-tiwari

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Abstract Views
PlumX Metrics