Zero Intelligence Models of the Continuous Double Auction: Econometrics, Empirical Evidence and Generalization

TL;DR

This paper develops a statistical method to estimate zero intelligence models of the continuous double auction using high-frequency quote data, identifies their limitations near the ask price, and proposes a generalized model that improves fit across three US markets.

Contribution

It introduces a generalized zero intelligence model that better captures limit order book behavior near the ask price, validated with empirical data from multiple markets.

Findings

Existing models fail near the ask price

The generalized model fits empirical data better

Significant improvements over previous models

Abstract

In the paper, a statistical procedure for estimating the parameters of zero intelligence models by means of tick-by-tick quote (L1) data is proposed. A large class of existing zero intelligence models is reviewed. It is shown that all those models fail to describe the actual behavior of limit order books close to the ask price. A generalized model, accommodating the discrepancies found, is proposed and shown to give significant results for L1 data from three US electronic markets. It is also demonstrated that the generalized model preforms significantly better than the reviewed models.