Reinforcement Learning in Repeated Portfolio Decisions

How do people make investment decisions when they receive outcome feedback? We examinedhow well the standard mean-variance model and two reinforcement models predict people’sportfolio decisions. The basic reinforcement model predicts a learning process that relies solelyon the portfolio’s overall return, whereas the proposed extended reinforcement model also takesthe risk and covariance of the investments into account. The experimental results illustrate thatpeople reacted sensitively to different correlation structures of the investment alternatives, whichwas best predicted by the extended reinforcement model. The results illustrate that simplereinforcement learning is sufficient to detect correlation between investments.