Machine learning the performance of hedge fund

Abstract

This study utilizes generative AI to predict and classify the performance of hedge funds based on groups of fund characteristics. Compared to commonly used machine learning methods, our method can successfully distinguish high- and low-performing funds across various investment strategies, with the return spread being the highest in the equity hedge strategy at 3.16 % monthly. The results are robust in risk-adjusted return prediction. Trend-based features are the most important predictors of future fund performance. Returns of predictive long-short portfolios are higher following periods of low narrative attention and favorable macroeconomic conditions. The asset allocation exercise highlights the significant economic value of machine learning. Our study enriches the burgeoning field of machine learning and artificial intelligence for finance by applying big data techniques to fund selection and allocation.