Bridging Temporal Dependencies and Sentiment: A Comprehensive Approach to NIFTY 50 Index Prediction

The primary goal of this research is to enhance the predictive accuracy of stock market trends by developing a comprehensive model that integrates Long Short-Term Memory (LSTM) networks with BERT-based sentiment analysis. This study aims to address the gap in traditional prediction models by exploring whether integrating sentiment from financial news with temporal market data can significantly improve forecasting of the NIFTY 50 index, a key benchmark in the Indian stock market. The proposed approach leverages critical financial indicators such as Foreign Institutional Investor (FII) and Domestic Institutional Investor (DII) activities, the India VIX (Volatility Index), and the Put-Call Ratio (PCR) of the two nearest expiry options. LSTM networks are employed to capture the temporal dependencies in historical market data, while BERT is used to extract sentiment insights from news articles. The unique contribution of this research is a dual-perspective model that combines quantitative financial data with qualitative sentiment analysis, which significantly outperforms traditional models in predicting bullish, bearish, or neutral market trends. This study provides a valuable tool for retail investors by offering more nuanced and accurate forecasts, underscoring the importance of integrating diverse data sources in stock market prediction. The findings suggest potential applications for similar predictive models in other financial markets.