Risk aversion impact on investment strategy performance: A multi agent-based analysis

Abstract

In order to supply an additional evidence on the effect of individual investors preferences on their portfolio dynamics from the wealth and risk adjusted return point of view, we construct an agent-based multi-asset model. We populate the artificial market with heterogeneous mean-variance traders with quadratic utility function. We compare the relative performance of investment strategies differ on their risk preferences using ecological competitions, where populations of artificial investors co-evolve. Our findings show that the higher relative risk aversion helps the agents survive in a long-range time frame in the competitions for higher wealth or Sharpe ratio of constrained portfolios. However, when short-selling is allowed, the highest (as well as lowest) risk aversion does not guarantee the highest earnings. Risk lovers as well as absolute risk averse run quickly out of competitions. Only the traders with moderate level of risk aversion survive in the long run. © Springer-Verlag Berlin Heidelberg 2012.