Predicting the change of stock market index based on social media analysis

Abstract

Stock market is definitely one of the common choices for investment throughout the world. Stock market index is well known as the composite of several representative stocks and reflects the trends of future stock market. Predicting the change of stock market index becomes crucial for individual investor, companies and stock holders. Recent research from behavioral finance implies that emotions of investors from social media can also influence stock market index in addition to the commonly used financial factors. Taking the advantages from the development of modern network and the age of big data, we are able to obtain information from different sources. More specifically, accessing user data from social media is no longer a challenge. Thus we apply data mining methods and propose to use the key words of emotions of market participates obtained from social media through text mining to predict the change of stock market index. Compared with traditional methods, we are able to utilize full information from the social network. We apply the propose approach in a dataset collected from Xueqiu forum, and the results show that linear discriminant analysis could give relatively good predictions of stock market index.