Reinforcement Learning in Experimental Asset Markets

Abstract

In this paper, we study the learning behavior possibly emerging in six series of prediction market experiments. We first find, from the experimental outcomes, that there is a general positive correlation between subjects’ earning performance and their reliance on using limit orders to trade. We therefore focus on the subjects’ learning behavior in terms of their use of limit orders or market orders by estimating a three-parameter Roth–Erev reinforcement learning model. The results of the estimated parameters show not just their great heterogeneity, but also the sharp contrasts among subjects, which in turn impact the subjects’ earning performance.