Using Financial News Sentiment for Stock Price Direction Prediction

Using sentiment information in the analysis of financial markets has attracted much attention. Natural language processing methods can be used to extract market from texts such as news articles. The objective this paper is articles and use estimated scores predict price direction stock index Standard & Poor’s 500. To achieve best possible performance classification, state-of-the-art bidirectional encoder representations transformers (BERT) models are used. pretrained transformer networks fine-tuned on a labeled text dataset applied known providers content their scores. generated for titles given articles, (text) said combined title-content consideration posited against past time series index. forecast index, predicted simple strategy features random forest classifier. results show that based particularly useful prediction.