Price predictability in limit order book with deep learning model

TL;DR

This paper investigates high-frequency price change prediction using deep learning, highlighting the importance of target process definition and volume imbalance for improving directional prediction accuracy.

Contribution

It demonstrates that proper target process definition and volume imbalance features enhance deep learning-based price prediction in limit order books.

Findings

Volume imbalance improves directional prediction performance.

Inadequate target process definition can make predictions meaningless.

Deep learning models perform well but are complex and less interpretable.

Abstract

This study explores the prediction of high-frequency price changes using deep learning models. Although state-of-the-art methods perform well, their complexity impedes the understanding of successful predictions. We found that an inadequately defined target price process may render predictions meaningless by incorporating past information. The commonly used three-class problem in asset price prediction can generally be divided into volatility and directional prediction. When relying solely on the price process, directional prediction performance is not substantial. However, volume imbalance improves directional prediction performance.