Model-Free Inference of Investor Preferences: A Relative Entropy IRL Approach

Chen Xu

We present a framework using Relative Entropy Inverse Reinforcement Learning (RE-IRL) to recover investor reward functions from observed investment actions and market conditions. U…