A multi-agent nonlinear Markov model of the order book

We introduce and treat rigorously a new multi-agent model of the order book (OB). Our model is designed to explain the collective behavior of the market when new information a?ecting the market arrives. Our model has two major features. First, it has two additional parameters which we call slow vari- ables. These parameters measure the mood of two groups of investors, namely, bulls and bears. Second, our model captures the interaction between trading agents and constitutes a nonlinear Markov process which exhibits long term cor- relations. We explain the intuition behind the equations and present numerical simulations which show that the behavior of our model is similar to the behavior of the real market.