Sentiment Analysis for Stock Price Flow Analysis

Abstract

Social media comments and news on Internet regarding any company can impact the their flow of stock prices. Often, if a company has bad review about their product in social media can have impact on their stock prices a lot. In this project, we investigate the relationship between StockTwits messages relationship with stock market movement. Specifically, we wish to see if, and how well, sentiment extracted from these feeds can be related to the shifts in stock prices. For this case we chose Apple Inc to perform the analysis. To answer this question, we construct a model, estimate its accuracy, and put it to the test on real market data. The state of the art in sentiment analysis suggests there are 2 important mood states that enable the prediction of mood in the general public. The prediction of mood uses the sentiment word lists obtained in various sources where general state of mood can be found using such word list or emotion tokens. With the number of messages posted on StockTwits, it is believed that the general state of mood can be predicted with certain statistical significance.