Sixteenths or pennies? Observations from a simulation of the Nasdaq stock market

Abstract

We have built a model that represents a highly realistic picture of a dealer-mediated market like Nasdaq, with the flexibility to model many features of real-world markets. While we have conducted a fairly significant amount of research using the model, we have limited it to four areas: 1) investigating, mainly in a qualitative fashion, the consequences of regulatory and structural changes to the market (the most important being tick size effects); 2) investigating whether our model, at least in a stylized fashion, is able to replicate some of the observed features of real-world markets; 3) validating the model (this encompasses the previous two points); 4) designing learning agents, and investigating the behaviors they learn and their ability to perform profitably in the market. Our results are significant in two respects. First, the model is robust: the simulated market as a whole, as well as the investors and dealers that make it up, perform realistically under a wide variety of conditions. Second, the market dynamics produced by the model have the same qualitative properties as those observed in real markets. Thus the model provides a test bed in which to investigate the effects of changes in market rules and conditions, and to investigate other aspects of the Nasdaq market.