Micro-price trading in an order-driven market

Limit order book simulations based on “zero-intelligence” or “entropy-maximizing” agents address two difficult issues in financial economics. First, the models address the significance of trading mechanisms by explicitly accounting for the logic of those mechanisms. Second, they avoid the difficulty of modeling human decision-making by generating orders stochastically. This paper reports on a computational experiment in which a strategic agent trading on endogenous market signals is embedded in an otherwise stochastic order book simulation. Under certain parameterizations of the model the agent is profitable despite the fact that the agent only employs market orders.