Modelling high-frequency limit order book dynamics with support vector machines

Abstract

We propose a machine learning framework to capture the dynamics of high-frequency limit order books in financial equity markets and automate real-time prediction of metrics such as mid-price movement and price spread crossing. By characterizing each entry in a limit order book with a vector of attributes such as price and volume at different levels, the proposed framework builds a learning model for each metric with the help of multi-class support vector machines. Experiments with real data establish that features selected by the proposed framework are effective for short-term price movement forecasts.

Keywords:

• High-frequency limit order book
• Multi-class classifiers
• Support vector machines (SVMs)
• Machine learning

JEL Classifications:

• 93E10
• 93E14
• 60G35
• 60H07

Acknowledgements

The authors would like to thank the Chair of Econometrics at Humboldt-Universität zu Berlin, Germany, for providing data used in this paper.

Notes

No potential conflict of interest was reported by the authors.

1 This optimization problem is solved experimentally using a JAVA implementation of the Sequential Minimal Optimization algorithm (Platt 1999).