Stock Price Prediction Using Sentiment Analysis on Financial News

Abstract

Owing to the tampered variance in the growth of the companies, it’s obvious to observe an extreme volatility in the stock values of the same. It is difficult to forecast a stock value since they depend on a variety of sociopolitical and economic circumstances, a change in leadership, investor attitude, and several other reasons. There is a significant association between the stock prices movements and the news release of a particular stock or related stocks, according to previous research in sentiment analysis. There might be a positive review or a negative review on the company’s technicals or finances based on which the investing pattern changes significantly. Previous approaches to predicting stock values using sentiment analysis have made use of algorithms like Support Vector Machine (SVM), Naive Bayes regression, and deep learning. Since the textual information gathered and examined was insufficient, the predictions turned out to be inaccurate. This paper creates a user facing app where the user can choose a particular stock among the listed stocks in the stock market, and the app will perform a news article hunt for the desired stock. Based on the information in the news articles, the app will perform an analysis and then predict a precise value or price to buy or sell stocks along with a preferred volume. In this project, we use deep learning models to collect a lot of time series data, train and test, and apply sentiment analysis in order to increase the accuracy of stock price forecasts. The dataset we compile dynamically comprises daily stock values for any chosen firm or company through its relevant financial news articles, as well as past technical records. A crucial part of our platform which predicts models and makes inferences for a stock in real time is cloud computing. Cloud computing helps us in deploying the software in the cloud, enabling global usage of the application. abstract environment.