Agent-based simulation of stock price in an artificial stock market

Abstract

The world financial markets and systems had crashed or failed several times in the previous and current century due to investors’ behaviours and attitudes as well as erratic developments on the market that arose out of such trading behaviours. The most notable examples include Wall Street crash of 1929, the dot com bubble burst and the recent US housing bubble.

In the real world where abstract complex systems has to be analysed, tested and debugged, it is often difficult, expensive and/or constrained by projects’ schedules to build the actual system or model for such purposes. Thus, computer simulation is often the favored approach taken by many researchers and engineers to deal with such problems.

In an attempt to investigate and further understand how investors’ behaviours affect stock price dynamics which lead to bullish and bearish markets, multi-agent-based computer simulation can be utilised to model the investors and the artificial stock market entities for such studies. Agent-based simulation approach also provides the benefit of allowing inter-agent social interactivity in the aspect of coordination, negotiation and communication. These social characteristics of agents can be used to imitate that of the actual behaviours of traders in the real world of trading and finance.

This report details the research done using JADE (as the agent simulation package) to implement investors’ trading mentality based on Day and Huang(1990) total excess demand macroeconomic theory and the corresponding consequences on the stock price that arise out of different trading behaviours. The observed phenomenon and data generated from the simulation model are then compared and contrasted against the actual data obtained from Asian markets such as the Straits Times Index (STI), the Hang Seng Index (HSI) and the Nikkei Index.