Modeling the bid/ask spread: measuring the inventory-holding premium

Abstract

We examine the relationship between the business cycle, sentiment, and the returns of listed U.S. hedge funds. Using Natural Language Processing (NLP) techniques, we construct a novel measure of hedge fund sentiment by mapping fund-level sentiment scores to hand-collected portfolio manager commentaries. Our empirical analysis shows that business cycle fluctuations exert the strongest influence on hedge fund sentiment, outweighing the effects of geopolitical, trade, and climate policy risks. Moreover, hedge fund sentiment exhibits explanatory power for the cross-section of returns, where a one-unit improvement in sentiment (from neutral to positive) is associated with an average annual return increase of approximately 0.74 percentage points.