Stock Price Prediction Using Sentiment Analysis of News Articles

Abstract

In the finance domain, the stock market exhibits inherent volatility, captivating researchers eager to decipher and anticipate its unpredictable patterns. Investors and market analysts diligently scrutinize market behavior, devising strategic buy or sell decisions. Given the vast daily data churned out by the stock market, individuals face a daunting task sifting through current and historical information to prognosticate future stock trends. Two primary methodologies for forecasting market trends are technical and fundamental analysis.Technical analysis relies on past price and volume data to forecast future trends, while fundamental analysis involves delving into a business's financial data for insights. The effectiveness of both these analyses is contested by the efficient-market hypothesis, positing that stock market prices are fundamentally unpredictable. This study, however, adopts the fundamental analysis approach, centering on news articles as pivotal information sources. It endeavors to classify news sentiment as either positive or negative, correlating positive sentiment with potential stock price increases and negative sentiment with potential decreases.The research strives to construct a model predicting news polarity and its potential impact on stock trends, essentially assessing the influence of news articles on stock prices. Employing supervised machine learning and text mining techniques, the study uses three distinct classification algorithms to enhance accuracy and classify unknown news articles. Three models, leveraging Random Forest, Support Vector Machine (SVM), and Naïve Bayes, demonstrate consistent performance across various testing scenarios. While Decision Tree yields satisfactory results, it falls short compared to the other models.The experimental phase validates the proposed model's efficacy, with results consistently surpassing an 80% accuracy rate. This research contributes valuable insights into the relationship between news sentiment and stock price movements, utilizing a robust methodology that combines fundamental analysis and machine learning techniques. © 2023 IEEE.