Stock Trend Prediction using Historical Data and Financial Online News

Financial online news on social networks has been proven to be a crucial factor that causes fluctuations in stock market. Regarding the impact of financial online news, this paper introduces Convolutional Neural Network (CNN) and Bi-directional Long Short-Term Memory (Bi-LSTM) to exploit the relation between financial online news and the fluctuations of stock price. A stock trends prediction model is proposed by deep combining the historical financial data feature, the news event feature and the sentiment orientation feature. News events and the corresponding sentiment orientations of financial news are introduced to help improving the accuracy of stock trends prediction. In order to verify the applicability of this model on different industries to predict the trend of individual stocks, two stocks are selected as the experimental objects, i.e., GREE Electric Appliances in the household appliance industry, and ZTE in the electronic appliance industry. The experiment results conducted on the past ten years data show that the proposed model improves the prediction accuracy about 10% and 20% in the household appliance industry and the electronic appliance industry, respectively, compared with the baseline algorithms.