A Deep Learning Model for Stock Price Prediction in Swing Trading

Abstract

As deep learning emerges and achieves remarkable success in many application areas, this paper presents a deep learning model for stock price prediction based on Multi-Input LSTM (MI-LSTM). In addition to new neural network architecture, we also try to take advantage of human traders’ wisdom by including the values of some recognized technical indicators in the network input in addition to raw prices. Experimental results show that our model could achieve more than 10% loss reduction, promising in higher potential trading profits.