Learning and the Price Dynamics of a Double-Auction Financial Market with Portfolio Traders

In this paper we study the dynamics of price adjustments in an artificial market where portfolio traders with bounded rationality and limited resources interact through a continuous, electronic open book. The present work extends the model developed in [? ] introducing endogenous target individual portfolio holdings. We model the agents’ order-flow investment decision as an optimal choice given individual characteristics and the available information. We depart from the standard asset pricing framework in two ways. First, we assume that investors have imperfect information about the returns distribution. In particular, we assume that agents hold arbitrary priors about securities’ returns, while they share a common constant view of the securities association structure. Agents properly update their univariate marginal distributions using historical data, as well as they derive, by appropriate use of the copula approach, the multivariate returns distribution to be used to find the optimal portfolio holdings. Thus, we concentrate our attention on analyzing the impact on price changes of uncertainty about the univariate marginals, allowing agents to have heterogeneous views about the shape of those distributions. Second, we use a prospect-type utility function to model the portfolio choice problem. Each investor has an initial level of wealth and a target growth rate to reach within his investment horizon. The investor must determine an asset allocation strategy so that the portfolio growth rate will be sufficient to reach the target. We model the utility function in terms of deviations from a specified target, and we assume that investors are more sensitive to downside movements. We run a series of simulations to answer the following questions: How does the learning process influence short-run and long run market results? How are the portfolio choices affected by the learning process?