Trading Decision Making Based on Hybrid Neural Network

Abstract

Now, with the dominance of electronic stock trading, it is possible to find and make profit from the price difference in real time. Machine learning has been applied in stock trading for years by companies. Yet as the rising of deep learning, price forecasting models become more accurate, which create more opportunities to gain higher profits. In this work, a novel hybrid neural network is proposed to deal with the stock trading decision making challenge. After properly training with labeled stock trading data, the hybrid neural network model proposed in this paper has been proved to be able to assist stock trading decisions better and achieve higher profits. The proposed hybrid neural network is evaluated on the stock trading data of Jane Street data set provided by the Kaggle competition. It is shown in the experiments that the proposed hybrid neural network outperforms other neural networks. Our neural network achieves as high profit value as 11417, which reveals the efficiency of the proposed method.