Agent-based modeling to investigate the disposition effect in financial markets

One of the behavioral patterns that deviate from what is predicted by traditional financial theories is the disposition effect. Although most empirical studies have reported a significant disposition effect, researchers have yet to conduct a conclusive test of thiseffect because a competing hypothesis or confounding effects might explain the documented significance. Thus, we use the tools of computational intelligence, instead of empirical approaches, to explore market behavior. In particular, we allow agents with different investment strategies to interact and to compete with each other in an artificial futures market. We found that the S-shaped value curve proposed by prospect theory may be one of the causes of the observed behavior of the disposition effect. However, rational expectation such as short-term mean reversion can even be more decisive