Deep Learning for Stock Market Prediction Using Social Media and Technical Information
Abstract— In recent years, the stock market has played an increasingly important role and attracted more and more attention. However, the complexity of the stock market makes stock prediction facing a considerable challenge. Many studies found that investor sentiment and stock technical indicators have a secure connection with the stock market movement. Also, in recent studies, deep learning has been widely used in time series forecasting and natural language processing, making it possible to predict stock markets successfully. In this paper, we apply the two-layer bidirectional long short-term Memory networks(Bi-LSTM) model based on glove word embedding and attention mechanism to extract stock sentiment indicators from social media, and use decision tree (DT) and principal component analysis (PCA) integrated model to extract stock technical indicators; then, these indicators apply to the LSTM model to forecast the US stock market movement. The experimental results show that our proposed method can significantly improve the accuracy of the stock market forecast.
Index Terms— Stock market; Deep learning; Social media; Time series forecasting; Technical indicators; Attention mechanism; Glove word embedding; LSTM; Bi-LSTM
Di Wu, Jianhua Cao
School of Computer Science and Technology, Dalian University of Technology, CHINA
Cite: Di Wu, Jianhua Cao, "Deep Learning for Stock Market Prediction Using Social Media and Technical Information," Proceedings of 2019 the 9th International Workshop on Computer Science and Engineering, pp. 88-98, Hong Kong, 15-17 June, 2019.