Predicting Stock Market Movements Through Daily News Headlines Sentiment Analysis: US Stock Market

This study leverages features extracted from daily world news headlines between August 2008 and June 2016 to predict next day’s Dow Jones Industrial Average (DJIA) index movement. Other related index return, commodity price changes and trading volume data are used to improve the prediction accuracy. In this work, the predictive models are based on three machine learnings algorithms: Random Forest, Support Vector Machine and Naïve Bayes. Based on the prediction results of three predictive models, a majority vote is also conducted. The results show that: (1) general daily world news headlines are less correlated with next day’s Dow Jones Industrial Average index movement; (2) sentiment implied in the news with indicators from technical analysis such as past return or trading volume can help to improve the prediction performance; (3) Naïve Bayes shows the superior ability in predicting next day’s index change, while majority voting of different predictive models can further the prediction accuracy.