Improvement methods for stock market prediction using financial news articles

Abstract

News articles serve the purpose of spreading company's information to the investors either consciously or unconsciously in their trading strategies on the stock market. Because of the immense growth of the internet in the last decade, the amount of financial articles have experienced a significant growth. It is important to analyze the information as fast as possible so they can support the investors in making the smart trading decisions before the market has had time to adjust itself to the effect of the information. This paper proposes an approach of using time series analysis and improved text mining techniques to predict daily stock market directions. Experiment results show that our system achieved high accuracy (up to 73%) in predicting the stock trends.