Using Deep Learning for price prediction by exploiting stationary limit order book features

The recent surge in Deep Learning (DL) research of the past decade has successfully provided solution to many difficult problems. The field of Quantitative analysis has been slowly adapting the new methods to its problems, but due to problems such as the non-stationary nature of financial data, significant challenges must be overcome before DL is fully utilized. In this work a new method to construct stationary features is proposed such that allows DL models to be applied effectively. These features are thoroughly tested on the task of predicting mid price movements of the Limit Order Book. Several DL models are evaluated such as recurrent Long Short Term Memory (LSTM) networks and Convolutional Neural Networks (CNN). Finally a novel model that combines the ability of the CNN to extract useful features and the ability of LSTMs’ to analyse time series, is proposed and evaluated. The combined model is able to outperform the individual LSTM and CNN models in the prediction horizons that are tested. •Limit Order Book data is presented and ways to analyse it with Deep Learning models.•Stationary features improve performance when compared to raw features.•Several Machine Learning models are compared for predicting price movements.•A combination of CNN and LSTM model is presented with improved performance.