Abstract:
The notion of stock market gains is an enticing one. For researchers, however, succeeding in developing a system that can predict market movements can be in and of itself an even more rewarding feat. With the rise of arti cial intelligence in general, and machine learning-based sentiment analysis in particular, the dream of stock market prediction has never been closer to our grasp. By leveraging the massive amounts of news data being cranked out daily, we can gauge the market mood via sentiment analysis techniques. We develop a novel version of the random
forest classi er infused with the powers of collocation and concordance, both of which borrowed from the eld of linguistics. Our experimental analysis yields insightful and impressive results compared to other works in the literature. Our novel model achieves a whopping 85% accuracy in predicting stock movements.
