As per capita incomes rise, more investors engage in high-risk, high-return stock investments. Market uncertainties significantly affect investor outcomes, making accurate stock price forecasting critically important. This study proposes a hybrid forecasting model combining Autoregressive Integrated Moving Average (ARIMA) and Long Short-Term Memory (LSTM) networks enhanced by sentiment analysis of publicly available financial advisor commentary. By extracting the emotional tone from advisor-level textual posts through the advanced natural language processing, the sentiment scores are integrated with traditional market data to improve the prediction of the National Stock Exchange (NSE) closing prices. This combined approach captures both linear market trends and nonlinear sentiment-driven fluctuations, overcoming the limitations of separate methodologies. The empirical results on the NSE indices highlight a significant relationship between the sentiment dynamics and the stock price variation. The hybrid ARIMA and LSTM model incorporating macroeconomic variables and sentiment indicators demonstrates improved forecasting accuracy, aiding investors in risk reduction and decision-making. This research advances the understanding of the behavioural influences on the financial markets and offers a practical tool for evidence-based trading strategies in the volatile emerging markets.