Extending Deep Learning Models for Limit Order Books to Quantile Regression

TL;DR

This paper introduces a deep learning approach using Quantile Regression to forecast financial returns from Limit Order Book data, enhancing prediction robustness and accuracy across multiple instruments.

Contribution

The paper develops a novel deep learning architecture that models return quantiles for buy and sell positions simultaneously, applied to high-frequency LOB data.

Findings

Model achieves high accuracy in predicting return quantiles.

Quantile estimates improve robustness of financial forecasts.

Method tested on millions of LOB updates across various instruments.

Abstract

We showcase how Quantile Regression (QR) can be applied to forecast financial returns using Limit Order Books (LOBs), the canonical data source of high-frequency financial time-series. We develop a deep learning architecture that simultaneously models the return quantiles for both buy and sell positions. We test our model over millions of LOB updates across multiple different instruments on the London Stock Exchange. Our results suggest that the proposed network not only delivers excellent performance but also provides improved prediction robustness by combining quantile estimates.