Simulação de mercado financeiro com compra e venda otimizadas por Enxame de Partículas

Abstract

It has been of great interest, both on the part of researchers and investors, to define negotiation rules that make it possible to capture the dynamics of the financial markets. This article presents a negotiation model among financial agents, based on the stock buying and selling process, that form a financial market. For the adaptation of economic agents to the market, a Particle Swarm Optimization (PSO) algorithm was implemented to optimize trading rules between agents aiming at maximizing gains in the market. Times series of artificial markets and real Bovespa brazillian market, descripted by the index Bovespa, were used in the computational simulations. Through the simulations, the dynamics of the interaction of buying and selling between financial agents was explored. The results show that there is a dependence on the gains of the agents in the markets in relation to the trading strategies adopted. On the other hand, in the low markets this dependence was not observed, since no statistically significant differences were found for the amount of wealth accumulated among the market participants. For the Bovespa market, from the sell and purchase thresholds of the trades carried out, the agents that have the best strategies in the negotiations were identified.