Sentiment Analysis of Stock Blog Network Communities for Prediction of Stock Price Trends

Abstract

It has been a challenge to develop a successful model for accurate stock price trend prediction. This paper aimed to develop an accurate model based on semantic analysis of social network communities for predicting stock day end closing prices using the wisdom of crowds ; www.mmb.moneycontrol.com, a financial blog covers all the companies listed on the National Stock Exchange of India. Influential and accurate opinions expressed in the blogs lead to community formation. The study focused on detection of such communities using betweenness centrality measure and performed a semantic analysis of their content to develop a prediction model based on correlation between blog sentiments and stock day end closing prices for predicting the stock trends. Thirty nine Indian banks were selected for the study during the period from October 1, 2017 to December 31, 2017 and the experimental results of the number of correct predictions of upside and downside movement of day end stock price were validated against the actual values. The model achieved a prediction accuracy of 84% and correlation of the model was within the significant limits of Z - test and Pearson's coefficient of 0.8.