Price formations by Monte-Carlo method: a simple comparison

This paper studies similarities and differences between models based on Monte-Carlo method by focusing on so-called “stylized facts”. In this study, we propose a model based on evolutionary algorithm and take the other model based on statistical physics. For this purpose, first we present a genetic learning model of investor sentiment and then several ordinary time series analyses are conducted after generating sample paths. Finally, the price properties are compared to those in the Ising spin model using the same algorithm. Our results show that both the Monte-Carlo simulations seem to lead to the similar dynamics reported in real markets in that the agents are boundedly rational or have some biases towards to the market. However, other time series properties are apparently different since the algorithm of price formation is different.