A Framework for Multi-agent-Based Stock Market Simulation on Parallel Environment

Multi-agent-based modeling is an efficient way to simulate the stock market which can be regarded as a complex adaptive system. This paper introduces a framework to support massive amounts of agents in the stock market simulation on distributed parallel environment. The framework consists primarily of the market server and massive amounts of agents implemented base on Erlang language. It provides an interface for decision mechanism of agents and interface for auction rules of market server. The experiment is established in a distributed environment and presents a certain scalability of the framework.