Enhancing Stock Market Forecasting: A Hybrid Machine Learning Approach Integrating LSTM and GRU Models

In this study, we address the limitations of traditional methods for stock price prediction by introducing an innovative hybrid machine learning model that integrates Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) networks. This model strategically layers LSTM and GRU modules with dropout regularization to harness both long and short-term dependencies in the data, concluding with a dense layer network that fine-tunes the prediction before arriving at the output layer. Utilizing market data from Yahoo Finance spanning from 2014 to 2024, the model was tested on significant market players including Tesla (TSLA), Apple (AAPL), Amazon (AMZN), and the S&P 500 index. The hybrid LSTM-GRU model excels in accuracy, outstripping standalone models in stock price forecasts. Metrics for Tesla showed RMSE at 0.1762 and MAPE at 0.0178%, Apple had RMSE of 0.1657 and MAPE of 0.0171%, Amazon recorded RMSE of 0.1892 with MAPE of 0.0191%, and the S&P 500 index exhibited RMSE of 0.2440 with a MAPE of 0.0125%, confirming the model's superior predictive performance.