Stock Exchange Prediction Using Financial News and Sentiment Analysis

Abstract

The main aim of this paper is to study the prominence of financial news on stock market prices and to put forward a new way to estimate stock prices employing news and sentiment analysis. Formerly, lot of attempts and research has been done to study the impact of historical stock market prices on the behavior of stock markets. Former outlook primarily focuses on either time series numerical data or textual documented text to presume the future variations of stock exchange trends which alone have proven to be insufficient. But with the onset of remarkable computing procedures and computing techniques like natural language processing, machine learning algorithms, etc., it becomes quite obligatory to view beyond just numerical data. This paper is one such bid and hence attempts to capture textual data as financial news to predict the recent market trends. There seems to be no correlation among stock market prices and news. This is due the fact that news is dynamic in nature and, therefore, cannot be used to forecast stock exchange prices. However, existing work suggests that the prosaic financial news have considerable and prominent ascendency on stock market. The concept behind implementing textual data is primarily based on fact that text inherently contains words that depict the sentiment of that text. The sentiment can either be positive or negative. Accordingly, the idea is to inspect the sentiment of the text to classify the text as being positive or negative with the presumption that positive news will affect markets positively while negative news will affect markets negatively. This paper poses a framework of predicting stock prices using sentiment analysis and financial news. The proposal also elucidates some sentiment analysis approaches.