Stock price prediction is useful for investment decisions but remains a difficult task because of fluctuations in stock prices and the presence of financial time dependencies. The traditional approaches fail to handle sequential patterns, which is why it is increasingly necessary to use AI methods. A deep learning-based financial forecasting system is suggested in this study. The system’s foundation is long short-term memory (LSTM) networks, which leverage the Yahoo Finance API’s real-time and historical stock market data. The dataset undergoes three data processing steps, including training-testing split at 80–20 ratios and normalization opposite to the resolution of the missing values. Eight samples per batch training support the LSTM model through ten epochs when three-layered cells (50, 100, 50 units) are used while the Adam Optimizer is employed along with the Mean Squared Error (MSE) and dropout (0.2) to avoid overfitting. The evaluation shows how LSTM outperforms alternative models, including KNN, by achieving 98.5% R2 and 0.572 RMSE values in root mean square error analysis. The LSTM model successfully detects market trends through predictive analysis of close and high prices, although it demonstrates some deviation during market volatility. The research demonstrates that LSTM executes successful pattern acquisition from financial time series, enabling it to operate as an excellent tool for stock market prediction and trend assessment while surpassing traditional machine learning techniques.