Forecasting price shocks with social attention and sentiment analysis

Many recent studies on finance and social networks discovered that investor's attention is correlated to the financial market movement in terms of the price shocks. Following related findings, a significant and challenging problem is to forecast the direction of the market movement based on vast social media activities. Appropriately processing social networks data and developing models to capture investor's attention on stocks would effectively help financial forecasting. In this paper, we propose and then apply a price shocks forecasting framework, which simultaneously takes the influence of social network users and their opinions about stocks into consideration. Specifically, we develop a new method to estimate social attention to stocks by influence modeling and sentiment analysis. Then, we use it in price shocks forecasting, which we formalize as a classification problem. We also consider the effect of historical market information on the market movement. Finally, we evaluate our framework based on a series of tests on the Chinese stock data. Our results show that the newly proposed measurement of social attention effectively improves the forecasting power of our framework.