Minority Game as a Model for the Artificial Financial Markets

What are the conditions required for a good artificial market? Simplicity, expandability, reality, good learning capability, sustainability are the elements definitely to be included into the model. Minority Game is one of those candidate models that depict necessary features for an artificial market. It is an evolutionary game designed for modeling the financial market made of heterogeneous agents with learning ability. Those agents are equipped with limited intelligence in space and time and they have access to public information, but independent otherwise. However, the rule of the original form of the minority game has an essential defect. The winners receive the same reward no matter how small the number of the minority agents is. This is unnatural, and this is the very reason why the conventional form of the minority game fails to reproduce the realistic wealth distribution among agents, namely, Gini´s coefficient being much too small compared to the observed value. To remedy this, winners with larger risk ought to be rewarded better. We show how this modification saves the model and reproduce the realistic wealth distribution ( G cong 0.5 ) known from the statistics.