Stock Price Prediction using Zero-Shot Sentiment Classification

Abstract

Investing in stocks is an important tool for modern financial management, wherein predicting the next day’s close price has become an issue. Social media, natural disasters, news, weather, and other such factors have an impact on stock market prices. The stock market can also become unpredictable at times. Sentiment Analysis (SA) combined with Machine Learning (ML) algorithms like Linear Regression (LR), Support Vector Regression (SVR), Naive Bayes, and Neural Networks was used by previous researchers to predict stock prices. Social media have more weight than other factors that influence stock price prediction. Accurately estimating a stock’s future price might produce substantial rewards. This research work has attempted to predict stock prices through Zero-Shot Sentiment Classification for Twitter Data and Deep Learning algorithms for stock price prediction. Historical stock data is also used to predict stock prices. Furthermore, the impact of Twitter data on stock prices is also discussed. The proposed model produces promising results with an accuracy of 94.21%.