A Bayesian learning model of hedge fund performance

We provide a Bayesian learning model in hedge fund performance. Our modelling provides a novel Bayesian aspirational model for panel data that is stable across different priors as reported from the mapping of the prior to the posterior of the Bayesian baseline model with the adoption of different priors. The parameters of our learning equation are time-varying which, to the best of our knowledge, is only addressed in Hu et al. (Strat Manag J 38:1435–1454, 2017) who assumed that the parameters have time and individual effects and depend on observed covariates. Our data set comes from the Lipper Trading Advisor Selection System database which includes data on performance and types of assets under management. Results reveal that a higher initial share price, management fee, leveraged and redemption notice period had a negative effect on learning. The learning curve has a U-shaped relationship, specifically, learning improves over the first three years, and gradually declines to zero by the eight-year. The second stage of analysis shows that though mean levels of learning do not directly influence performance, a higher standard deviation in learning lowers the decline in performance with higher mean learning. But, we report variability in results across various models that we test for robustness.