How to calibrate the features of a distributed artificial financial market

Abstract

The modelling of financial markets by making use of artificial financial agents is an analytical tool that has been highly developed during the past few years. The interaction of agents with different memories in specifying market prices results in the creation of a certain market that makes long-term prediction difficult. The structural characteristics of a computational-financial market that includes agents who are in the process of exchanging information and adapting to the conditions of the market over time are studied in this research. Characteristics of agents, together with the structural characteristics of the market, have been examined to calibrate the employed cellular learning automata framework. The results show that the calibration of these characteristics has the effect of simulating a market very similar to the real market. In this study, the market was simulated in a monthly period; however, the level of return in more distant time periods has been compared.