Systematic and Discretionary Hedge Funds: Classification and Performance Comparison

In this paper, we introduce an approach to building classifiers that bifurcate hedge funds into systematic and discretionary funds and evaluate their performance. This approach makes use of textual analysis and statistical learning methods that are free from the subjective judgment of investment strategies. In our empirical study, we find that a random forest classifier yields the highest accuracy ratio and that the funds classified as systematic, on average, result in higher excess returns than their discretionary counterparts. After controlling for the false discovery rate, we find that systematic funds are preferred to their discretionary ones because a higher portion of positive alpha~(skilled) funds are systematic funds. The skilled funds' alpha average is larger for systematic funds than for discretionary funds across all categories and variant observable and unobservable factor models we considered.