Trading Rule Search with Autoregressive Inference Agents

The use of the agent-based paradigm in modelling financial markets provides an intuitively natural approach and is a well established technique. In contrast with the assumptions and conclusions of the efficient markets hypothesis (EMH), agent based models provide a refreshing causal approach to understanding the emergence of the general stylized facts of financial markets. In this report we present details of an agent-based stock market simulation in which traders utilise a hybrid mixture of common information criteria based inference procedures, including minimum message length (MML) inference. Traders in our model compete with each other using a range of different inference techniques to infer the parameters and appropriate order of simple autoregressive (AR) models of stock price evolution. We show that in the presence of a noisy AR signal, MML traders significantly outperform their competitors, and in fact do well even in the absence of such a signal.