StTime-Net: Combining both Historical and Textual Factors for Stock Movement Prediction

With the development of modern technology, machine learning has become the most popular tool for analyzing financial and numerical data. It has great potential to forecast stock prices. The proper prediction of future stock prices helps investors to increase their chances of earning more profits. As a result, many research papers using machine learning and deep learning in stock price prediction have been conducted. However, most researchers only focus on the past price categories of the stock market without considering other essential types of information, which include the articles about a company and how people talk about a company on social media. Thus, in this paper, we apply machine learning and deep learning methods to predict the future stock market through both numerical data and textual information. The textual information is based on analyzing tweets about the companies. Moreover, we adopt Time2Vec [13] to learn a vector representation of time. We show that our method (StTime-Net) is a more suitable approach for stock movement prediction.