Stock Price Prediction Using Twitter Social Network Analysis

Over the past few years, social networks have become an important element for a huge amount of corporations. According to the efficient market hypothesis, social networks' data can improve standard methods of forecasting prices on stock markets. The aim of the research is to find and analyze the dependencies between Twitter's text tone about a company and the dynamics of prices using linear methods of time series correlation analysis and further price forecasting. The following major corporations were selected for the study: Amazon and Netflix. The time series analyzed is a set of hourly data that was obtained from Twitter and Finam websites. In the course of the research, linear methods of time series analysis proved the dependence between the dynamics of companies' prices and the mention of these companies in the social network. Also as a result of the study, future prices of companies were predicted using linear regression. Thus, we can say that the mood of people in the social network is a valuable and important source of information for the forecasting task.