Improving Stock Price Prediction Models Using Sentiment Analysis

With regards to behavioural financial theory, it is necessary to consider psychological aspects of investors when one wants to predict stock price. Everyday investors deal with various social media, blogs, news etc. Hence, these are appropriate resources for measuring investors sentiment leading to better prediction for what will happen in the future. The aim of this article is to use sentiment analysis and acquire sentiment index. Besides, we want to indicate the relationship between sentiment index and stock price movement by using LSDV regression. We select five companies from Dow Jones Industrial Average index and measure sentiment index. Then combine our sentiment index with historical data of price and volume of these five stocks for prediction. Results show that there is a significant relationship between stock price and mental status of investors. Also our propose model can define fluctuations of stock price by 91 percent. After that we prove that there is difference in relationship between each company individually and its own sentiment index.

Keywords:

Behavioral finance , Opinion mining , Sentiment analysis , Text mining