Deep Learning for Digital Asset Limit Order Books

9 Pages Posted: 20 Nov 2020

See all articles by Rakshit Jha

Rakshit Jha

University of Cambridge

Mattijs De Paepe

University of Cambridge

Samuel Holt

Independent

James West

Globe Research

Shaun Ng

Globe Research

Date Written: October 3, 2020

Abstract

This paper shows that temporal CNNs accurately predict bitcoin spot price movements from limit order book data. On a 2 second prediction time horizon we achieve 71% walk-forward accuracy on the popular cryptocurrency exchange coin-base. Our model can be trained in less than a day on commodity GPUs which could be installed into co-location centers allowing for model sync with existing faster order-book prediction models. We provide source code and data at https://github.com/Globe-Research/deep-order-book.

Keywords: Cryptocurrency, Bitcoin, CNN, Order-book, Market Making, Direction, Prediction, TCN, Temporal Convolutional Network

JEL Classification: G10, G11, G12

Suggested Citation

Jha, Rakshit and De Paepe, Mattijs and Holt, Samuel and West, James and Ng, Shaun, Deep Learning for Digital Asset Limit Order Books (October 3, 2020). Available at SSRN: https://ssrn.com/abstract=3704098 or http://dx.doi.org/10.2139/ssrn.3704098

Rakshit Jha

University of Cambridge ( email )

Trinity Ln
Cambridge, CB2 1TN
United Kingdom

Mattijs De Paepe

University of Cambridge ( email )

Trinity Ln
Cambridge, CB2 1TN
United Kingdom

James West

Globe Research ( email )

HOME PAGE: http://globedx.com

Shaun Ng

Globe Research ( email )

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
584
Abstract Views
1,563
Rank
72,996
PlumX Metrics