A Time-Series Forecasting Framework for Stock Market Prediction Using LSTM and Technical Indicators

This paper proposes a real-time framework of stock market prediction based on Long Short-Term Memory (LSTM) networks and prominent indicators. The first is to enhance the effectiveness of more accurate forecasting by measuring temporal dependencies and the non-linearity model in the data on finance. The Yahoo Finance API retrieves historical stock information augmented to indicators like the Relative Strength Index (RSI), Moving Averages (SMA, EMA), MACD, and Bollinger Bands. The model is an LSTM tested and compared against conventional machine algorithms such as Linear Regression, Support Vector Machines (SVM) and Random Forest. Performance evaluation is done through Mean Squared Error (MSE), Mean Absolute Error (MAE) and R 2 Score. The findings prove that the offered LSTM model demonstrates even better accuracy, with the value of MSE equal to 0.004 and R Sq Score reaching 0.95. By combining real-time information and deep learning, the interplay opens up an effective stock prediction model, a massive advancement compared to traditional forecasting systems.