Mining Textual Terms for Stock Market Prediction Analysis Using Financial News Asyraf Safwan Ab. Rahman, Shuzlina Abdul-Rahman (&) , and Sofianita Mutalib Faculty of Computer and Mathematical Sciences, Universiti Teknologi MARA, 40450 Shah Alam, Selangor, Malaysia asysafwan@gmail.com, {shuzlina,sofi}@tmsk.uitm.edu.my Abstract. This study focuses on the use of machine learning algorithms to construct a model that can predict the movements of Bursa Malaysia stock prices. In this research, we concentrate on linguistics terms from financial news that can contribute movements of the prices. Our aim is to develop a prototype that can classify sentiments towards financial news for investment decision. We experimented with five blue-chip companies from different industries of the top market constituents in Bursa Malaysia KLCI. A total of 14,992 finance articles were crawled and used as the dataset. Support Vector Machine algorithm was employed and the accuracy recorded was at 56%. The findings of this research can be used to assist investors in investment decision making. Keywords: Bursa Malaysia  Sentiment analysis  Stock market prediction Support Vector Machine  Text mining 1 Introduction Stock markets are where public companies issue or trade their shares on either exchanges or over-the-counter markets. Stock prices are determined by the supply and demand of a particular stock. Understandably, it can be argued that stock markets cannot be entirely predicted. Changes in these markets are often associated with the sentiments of an investor. In predicting stock markets, there are two analysis used, which are fundamental analysis and technical analysis. Fundamental analysis focuses on studying the business, competitors, and markets. Technical analysis, on the other hand, focuses on analyzing historical prices to determine the direction of stock prices [11]. Fundamental analysis is concerned with the availability of quantitative or qual- itative information to investors in driving investment decisions. News articles may carry useful information to market participants and by using fundamental analysis, an investment decision can be made. Basically, these articles usually come verbosely and lengthy in textual representation, thus extracting patterns manually can be time-consuming. There is a correlation between fluctuating stock prices and the publication of financial news [5]. Thus, changes in any stock market can be attributed to the publication of news. However, it is dif ficult to predict stock prices from textual data [12]. © Springer Nature Singapore Pte Ltd. 2017 A. Mohamed et al. (Eds.): SCDS 2017, CCIS 788, pp. 293–305, 2017. https://doi.org/10.1007/978-981-10-7242-0_25