Home / Current Issue / Paper 1712699
Stock Market Predictions Using Sentiment Analysis
Subject area: Management and Commerce · Area of research: Stock marketing
DOI: https://doi.org/10.64388/IREV9I6-1712699
Abstract
The prediction of stock prices is one of the significant areas of financial research which is mainly based on classical methodologies such as Technical Analysis and Fundamental Analysis . On the other hand, because of the increased use of the internet and the development of Social Media as a tool for communication, the new method of calculating investor sentiment regarding stocks has also emerged. This is known as "Sentiment Analysis." The method of calculating investor sentiment or stock sentiment, as it is referred to, combines both Natural Language Processing (NLP) and Machine Learning (ML) to collect data from social media, news articles or company earnings announcements, to calculate market sentiment about an individual company's stock price. The methodology for using Sentiment Analysis in this research includes the use of a Lexicon Baseline, Machine Learning Classifiers, and Deep Learning, including LSTM, BERT and FinBERT. The research assesses the capability of each methodology to predict stock prices based on the results of the sentiment scores combined with historical stock prices. The research indicates that deep learning methods are much better than traditional machine learning techniques with respect to predicting the movement of stocks based on sentiment. The top-performing model was FinBERT, which outperformed all other models in terms of accuracy predicting the sentiment of stock prices. The results also demonstrate that investment strategies that are based on sentiment produced superior returns and had significantly less volatility compared to traditional investment strategies that did not utilize sentiment. However, even with these promising results, sentiment analysis has many challenges, including data bias, the use of sarcasm, and venture capitalists' irrational behavior in the market. Future research can build on this work by creating tools that enable the real-time assessment of sentiment as well as hybrid models that integrate sentiment with macroeconomic indicators. Sentiment analysis, therefore, provides added value to investors by enhancing the accuracy of investment predictions.
Keywords
Stock Market prediction, Sentiment Analysis, Machine Learning, News Analytics, Text Mining, Algorithmic Trading, Machine Learning.
How to cite this paper
@article{1712699,
author = {Dr. Gurrapusala Venkata Siva, Dr. Arachana Nag },
title = {Stock Market Predictions Using Sentiment Analysis},
journal = {Iconic Research And Engineering Journals},
year = {2025},
volume = {9},
number = {6},
pages = {812-819},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1712699.pdf},
abstract = {The prediction of stock prices is one of the significant areas of financial research which is mainly based on classical methodologies such as Technical Analysis and Fundamental Analysis . On the other hand, because of the increased use of the internet and the development of Social Media as a tool for communication, the new method of calculating investor sentiment regarding stocks has also emerged. This is known as "Sentiment Analysis." The method of calculating investor sentiment or stock sentiment, as it is referred to, combines both Natural Language Processing (NLP) and Machine Learning (ML) to collect data from social media, news articles or company earnings announcements, to calculate market sentiment about an individual company's stock price. The methodology for using Sentiment Analysis in this research includes the use of a Lexicon Baseline, Machine Learning Classifiers, and Deep Learning, including LSTM, BERT and FinBERT. The research assesses the capability of each methodology to predict stock prices based on the results of the sentiment scores combined with historical stock prices. The research indicates that deep learning methods are much better than traditional machine learning techniques with respect to predicting the movement of stocks based on sentiment. The top-performing model was FinBERT, which outperformed all other models in terms of accuracy predicting the sentiment of stock prices. The results also demonstrate that investment strategies that are based on sentiment produced superior returns and had significantly less volatility compared to traditional investment strategies that did not utilize sentiment. However, even with these promising results, sentiment analysis has many challenges, including data bias, the use of sarcasm, and venture capitalists' irrational behavior in the market. Future research can build on this work by creating tools that enable the real-time assessment of sentiment as well as hybrid models that integrate sentiment with macroeconomic indicators. Sentiment analysis, therefore, provides added value to investors by enhancing the accuracy of investment predictions.},
keywords = {Stock Market prediction, Sentiment Analysis, Machine Learning, News Analytics, Text Mining, Algorithmic Trading, Machine Learning.},
month = {December},
doi = {https://doi.org/10.64388/IREV9I6-1712699}
}