Using SVM to Predict Stock Price Changes from Online Financial News

Many technical analysis use financial indices to predict stock price changes. In this paper, we present a different approach for prediction stock price fluctuations using financial news. Our method approaches the stock price prediction problem from an information retrieval perspective. We apply both text analysis and pattern classification techniques to search for important online news that are relevant for stock price changes. First, the online financial news and the corresponding stocks are extracted. Then we apply Support Vector Machine (SVM) to construct a model that predicts the price changes for the stocks. Finally, the stock changes prediction model is used to classify and extract upcoming important financial news. The experimental results demonstrate our method is effective for seeking the important financial news for stock price changes.