A multiagent system for efficient portfolio management

Abstract

In this work we present a multiagent system to draw up an optimum portfolio. By using a distributed architecture, the agents are trained to follow different investing strategies in order to optimize their portfolios to automate the one year forecast of a portfolio’s payoff and risk. The system allows to adopt a strategy that ensures a high rate of return at a minimum risk. The use of neural networks provides an interesting alternative decisions to the statistical classifier. With a modular agent composed by a few trained neural networks, the system makes investment decisions according to the assigned investment strategy and the behavior of the prices in a one-year period. The agent can take a decision on the purchase or sale of a given asset.