The stock market is a highly volatile and dynamic financial domain where predicting future price movements is an ongoing challenge for investors and financial institutions. With the advent of machine learning technologies, particularly deep learning models, the field of stock market prediction has gained significant momentum. In this study, we present a comprehensive approach to stock forecasting using a long short-term memory (LSTM) model. The proposed model leverages historical stock data, financial indicators, and sentiment analysis from news articles to make accurate predictions of stock prices. By utilizing a multivariate dataset that includes technical and fundamental market indicators, we aim to capture both short-term fluctuations and long-term trends in the price of shares. The novelty of this study lies in the combination of emotion analysis with traditional stock features, enabling the model to incorporate the influence of market sentiment on price movements. The dataset used includes daily stock prices from the S&P 500 Index, trading volumes, earnings reports, and sentiment data extracted from financial news. The proposed LSTM architecture is ideal for financial time-series forecasting since it is specifically made to manage sequential data and capture long-term dependencies. The methodology involves extensive feature engineering, including data normalization, correlation analysis, and denoising using wavelet transformation to enhance the quality of the input information. The findings show that the suggested LSTM model outperforms traditional methods in terms of accuracy. In summary, this research proposes a novel LSTM-based stock forecasting system that integrates historical stock data, financial indicators, and sentiment analysis to provide accurate and timely predictions. By addressing the challenges of nonlinear and complex financial data, the proposed system aims to empower investors with a reliable tool for portfolio optimization and risk management.