Integrating StockTwits with sentiment analysis for better prediction of stock price movement

Abstract

Sentiment Analysis is new way of machine learning to extract opinion orientation (positive, negative, neutral) from a text segment written for any product, organization, person or any other entity. Sentiment Analysis can be used to predict the mood of people that have impact on stock prices, therefore it can help in prediction of actual stock movement. In order to exploit the benefits of sentiment analysis in stock market industry we have performed sentiment analysis on tweets related to Apple products, which are extracted from StockTwits (a social networking site) from 2010 to 2017. Along with tweets, we have also used market index data which is extracted from Yahoo Finance for the same period. The sentiment score of a tweet is calculated by sentiment analysis of tweets through SVM. As a result each tweet is categorized as bullish or bearish. Then sentiment score and market data is used to build a SVM model to predict next day's stock movement. Results show that there is positive relation between people opinion and market data and proposed work has an accuracy of 76.65% in stock prediction.