A Hybrid Approach for Stock Market Prediction Using Financial News and Stocktwits

Stock market prediction is a difficult problem that has always attracted researchers from different domains. Recently, different studies using text mining and machine learning methods were proposed. However, the efficiency of these methods is still highly dependant on the retrieval of relevant information. In this paper, we investigate novel data sources (Stocktwits in combination with financial news) and we tackle the problem as a binary classification task (i.e., stock prices moving up or down). Furthermore, we use for that end a hybrid approach which consists of sentiment and event-based features. We find that the use of Stocktwits data systematically outperforms the sole use of price data to predict the close prices of 8 companies from the NASDAQ100. We conclude on what the limits of these novel data sources are and how they could be further investigated.