Deep learning for limit order books

By Justin A. Sirignano. 2016

137 citations recorded by the source corpus.

This paper develops a new neural network architecture for modeling spatial distributions (i.e. distributions on ) which is more computationally efficient than a traditional fully-connected feedforward architecture. The design of the architecture takes advantage of the specific structure of limit order books. The new architecture, which we refer to as a ‘spatial neural network’, yields a low-dimensional model of price movements deep into the limit order book, allowing more effective use of information from deep in the limit order book (i.e. many levels beyond the best bid and best ask). The spatial neural network models the joint distribution of the state of the limit order book at a future time conditional on the current state of the limit order book. The spatial neural network outperforms status quo models such as the naive empirical model, logistic regression (with nonlinear features), and a standard neural network architecture. Both neural networks strongly outperform the logistic regression model. Due to its more effective use of information deep in the limit order book, the spatial neural network especially outperforms the standard neural network in the tail of the distribution, which is important for risk management applications. The models are trained and tested on nearly 500 U.S. stocks. Techniques from deep learning such as dropout are employed to improve performance. Due to the significant computational challenges associated with the large amount of data, models are trained with a cluster of 50 GPUs.

Subfields: Finance, Management Science and Operations Research

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Finance

In Management Science and Operations Research

Citations in the corpus

Universal features of price formation in financial markets: perspectives from deep learning (2018, methodology): “This is one of the reasons why, except in a few instances [7, 14, 25, 28, 29, 30], large-scale learning methods such as Deep Learning [18] have not been deployed for quantitative modeling in finance.”
Tensor representation in high-frequency financial data for price change prediction (2017, methodology): “Popular machine learning methods in HFT include regression analysis [1], [2], [3], [4], [5], multilayer feed forward network [6], [7], [8], convolutional neural network [9] , recurrent neural network [10], [11], [12].”
Benchmark Dataset for Mid-Price Prediction of Limit Order Book data (2017, methodology): “On the other side, Sirignano [49] uses large amounts of data for nonlinear classiﬁcation based on feedforward network.”
Graph Element Networks: adaptive, structured computation and memory (2019, methodology): “…neural networks (Takeuchi & Kosugi, 1994; Ramuhalli et al., 2005) or Hopﬁeld networks (Guo et al., 1999; Al Salameh & Al Zuraiqi, 2008) as well as mesh-free methods to solve PDEs (Dissanayake & Phan-Thien, 1994; van Milligen et al., 1995; Leake, 2018; Long et al., 2018; Sirignano, 2018).”
Sequence classification of the limit order book using recurrent neural networks (2017, methodology): “A number of researchers have applied machine learning methods to the study of limit order book dynamics (Kearns and Nevmyvaka, 2013; Kercheval and Zhang, 2015; Sirignano, 2016; Dixon et al., 2017).”
A High Frequency Trade Execution Model for Supervised Learning (2017, unknown): “Following Kercheval and Zhang (2015); Sirignano (2016); Dixon (2017), we compose our feature set of ﬁve levels of prices, volumes and number of limit orders on both the ask and bid side of the book.”