Correlating and Predicting Stock Prices with Twitter Sentiments

Le résumé fourni par la source

This paper presents an empirical study of correlating Twitter sentiments with individual stock price movements. We used an existing text-mining technique, OpinionFinder, to extract Twitter sentiment data from plaintext tweets. Different from prior researches, we explored a novel approach to aggregate Twitter sentiment features and Twitter metadata features associated with the tweets that mention a technology stock to construct a set of features, which was then correlated with the stock price movements of the respective stock prices. We thereby selected a subset of these features, which have positive correlation coefficients with the stock prices, to predict future stock price movements. The results of the prediction, however, are not as successful as expected. Although it is too early to conclude that Twitter sentiments cannot be used to predict an individual stock price, our results do provide one piece of negative evidence for such hypothesis.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.