Social Media Based Stock Prediction

Abstract

The stock market is one of the most important part of the financial market. However, it is difficult for investors to extract information from data. In the meantime, sentiment analysis is a hot topic in computer science. In this paper, we build an emotion dictionary for finance area. We can calculate the emotional values of finance texts based on the dictionary. Further, with a regression analysis by Fama-French model, we discovered the relationship between the emotional value and stock price by experiments. Simply put, the emotional value and stock's return rate did not have a significant correlation; the emotional value and stock's volatility don't have a significant correlation; but the emotional value and stock' relative return rates had a positive correlation.