High-frequency financial data are characterized by a set of ubiquitous statistical properties that prevail with surprising uniformity. While these 'stylized facts' have been well-known for decades, attempts at their behavioral explanation have remained scarce. However, recently a new branch of simple stochastic models of interacting traders have been proposed that share many of the salient features of empirical data. These models draw some of their inspiration from the broader current of behavioral finance. However, their design is closer in spirit to models of multi-particle interaction in physics than to traditional asset-pricing models. This reflects a basic insight in the natural sciences that similar regularities like those observed in financial markets (denoted as 'scaling laws' in physics) can often be explained via the microscopic interactions of the constituent parts of a complex system. Since these emergent properties should be independent of the microscopic details of the system, this viewpoint advocates negligence of the details of the determination of individuals' market behavior and instead focuses on the study of a few plausible rules of behavior and the emergence of macroscopic statistical regularities in a market with a large ensemble of traders. This chapter will review the philosophy of this new approach, its various implementations, and its contribution to an explanation of the stylized facts in finance.
Speculation Stylized facts Group dynamics Agent-based models