Sequence Classification of the Limit Order Book Using Recurrent Neural Networks

Abstract

Recurrent neural networks (RNNs) are types of artificial neural networks (ANNs) that are well suited to forecasting and sequence classification. They have been applied extensively to forecasting univariate financial time series, however their application to high frequency trading has not been previously considered. This paper solves a sequence classification problem in which a short sequence of observations of limit order book depths and market orders is used to predict a next event price-flip. The capability to adjust quotes according to this prediction reduces the likelihood of adverse price selection. Our results demonstrate the ability of the RNN to capture the non-linear relationship between the near-term price-flips and a spatio-temporal representation of the limit order book. The RNN compares favorably with other classifiers, including a linear Kalman filter, using S&P500 E-mini futures level II data over the month of August 2016. Further results assess the effect of retraining the RNN daily and the sensitivity of the performance to trade latency.

An Empirical Study of Machine Learning Algorithms for Stock Daily Trading Strategy

“…In recent years, artificial intelligence computing methods represented by DNN have made a series of major breakthroughs in the fields of Natural Language Processing, image classification, voice translation, and so on. It is noteworthy that some DNN algorithms have been applied for time series prediction and quantitative trading [17][18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34]. However, most of the previous studies focused on the prediction of the stock index of major economies in the world ([2, 8, 11, 13, 15-17, 22, 29, 30, 32], etc.)…”

Learning from data streams using kernel least-mean-square with multiple kernel-sizes and adaptive step-size

“…Rather than throw away previously learned kernel-sizes, we use them to build a pool of kernel-sizes in an online sequential way. Then, for each new input sam- 4 Where: (1) the kernel-size at iteration t − 1 is σ t−1 ; (2) the free parameter for the kernelsize adaptation is ρ; (3) the prediction errors at time t − 1 and t are e t−1 and et, respectively;…”

“…The sequence imposes an order on the samples; this order must be preserved when training models and making predictions [1]. The goal being to extract knowledge from a continuous sequence of data records, e.g., financial markets, network traffic, weather conditions, among others [2,3,4].…”

Order flow analysis of cryptocurrency markets

“…Such data-driven approaches were especially made attractive with development of (a) fast and efficient computation, (b) vast amounts of data and (c) modern Machine Learning algorithms that include classical statistical learning techniques (Hastie et al 2001) and more recently, Deep Learning (Goodfellow et al 2016). Goals of such studies vary from studying market impact and order book modelling (Donier and Bonart 2015;Cont et al 2014), to extracting predictive capability from various market microstructure features (Dixon 2018;Sirignano and Cont 2018). Criticism around data-driven approach is centred around the fact that studies tackle the data head-on, often studying the statistical mechanics of the after-facts, rather than asking fundamental questions about possible origins of the underlying processes.…”