Stochastic beliefs learning and commodity futures price dynamics

We propose a multi-agent-based artificial futures market to understand commodity futures price dynamics, which is called stylized facts. The multi-agent-based artificial futures market includes hedger agents and speculative agents. We use Brenner's stochastic belief learning model (SBL) to describe speculative agents' learning process. The SBL is a social learning process with local information. New market price is generated though a sealed-bid auction clearance mechanism. Our simulation result can reproduce the important observed stylized facts of futures price. The simulation programming language is Matlab 7. Our results show clustered volatility depends on speculator agents' imitations which are caused by social learning process. Using this simulation model, we can find futures price volatility has close relation with large speculator agents' trading activity.