The stock market is an important part of a country’s economy because it shows the performance of companies and the overall business environment. Since stock prices change often and in unpredictable ways, it is very difficult to guess future trends. In recent years, artificial intelligence has become popular for this type of work. In particular, deep learning models such as LSTM (Long Short-Term Memory) are used for time series forecasting in finance. However, the success of LSTM models depends a lot on how they are designed and trained. In this study, we focus on building an improved LSTM-based framework for stock price prediction. To make the input data stronger, we also use Bollinger Bands. These help capture changes in price movements and market volatility, which makes the model’s learning more effective. The model is tested using historical data from the Dow Jones Industrial Average (DJIA). For comparison, we also check the performance against a simple ANN model and different LSTM variations. The evaluation is done using common error measures such as MSE, MAE, MAPE, and R². The results show that the LSTM model with Bollinger Bands gives better predictions than the other models. It provides higher accuracy and more reliable performance for forecasting stock prices.