Financial Markets and Trading: An Introduction to Market Microstructure and Trading Strategies

CHAPTER 9

Agent-Based Modeling of Financial Markets

The main idea behind the agent-based modeling is that agents' actions affect or even determine the environment in which they function. This framework is similar to the methodology of statistical physics where macroscopic properties of a continuum are calculated using averaging over inter-particle interactions. Agent-based modeling has become very popular in economics and finance (see the reviews by Hommes 2006, LeBaron 2006, and Chiarella et al. 2009).

The notion of agents is used in many market microstructure models discussed in former chapters (see, e.g., a review by Parlour & Seppi 2008). However, the two fields differ in their assumptions and in the object of research. Namely, the current market microstructure theory is focused primarily on the properties of the order book in equilibrium. It is usually assumed that all investors are rational and returns follow the random walk in the spirit of the EMH (see Chapter 7).¹ On the other hand, the main goal of agent-based market modeling is to describe how the asset price dynamics are affected by trader behavior. The deterministic component in agent-based models implies that financial markets are at least partially predictable, which is difficult to reconcile with the classical financial theory.

The agent-based models are capable of describing several stylized facts observed in financial markets. One of the first agent-based models of financial markets was offered by Beja & Goldman ...