Stochastic simulation framework for the Limit Order Book using liquidity motivated agents

TL;DR

This paper introduces a stochastic agent-based model for limit order books driven by liquidity-motivated agents, calibrated on real data, and useful for testing regulations and informing trading decisions.

Contribution

It presents a novel liquidity-driven agent-based modeling framework for limit order books with an efficient calibration method using real market data.

Findings

Successful calibration on Chi-X data

Framework useful for testing regulations

Informs brokerage and trading decisions

Abstract

In this paper we develop a new form of agent-based model for limit order books based on heterogeneous trading agents, whose motivations are liquidity driven. These agents are abstractions of real market participants, expressed in a stochastic model framework. We develop an efficient way to perform statistical calibration of the model parameters on Level 2 limit order book data from Chi-X, based on a combination of indirect inference and multi-objective optimisation. We then demonstrate how such an agent-based modelling framework can be of use in testing exchange regulations, as well as informing brokerage decisions and other trading based scenarios.