A Comparative Study into Stock Market Prediction Through Various Sentiment Analysis Algorithms

Abstract

From past times, there is strong link between finances, currencies and economic growth the future and progress of humankind. Now a days, the future of economic development is dictated by fortunes and vagaries prevalent in share stock markets. Researchers have found that it is possible to make forecasts with large historical data on stock market and up-down in prices of share values. So, the fact that stock markets play a really vital role in national and global economy, which is today undeniable. Stock markets can be profitable by speculations provided, of course though the future behavior can be forecast with a constant degree of accuracy. In this study, the authors propose a model which help to guess stock market trends consistently and with minimal of error value. The model discussed here makes the use of sentiment analyses based on financial news and also historical patterns of stocks in share markets and can offer more accurate results to analyses data from multiple news sources and historical price movement of individual stocks. By using a two-step process, the model offers a minimum prediction accuracy value of 72%. In the first step, Naïve Bayes algorithm is used to evaluate text polarity to obtain a fix on public sentiment based on news feeds collected and received. In the second step, the future stock prices are forecasted by combining the evaluation results on text polarity with historical data on stock value price up-down.