LSTM hyperparameter configuration.

With the increasing complexity and prosperity of global financial markets, stock market forecasting plays a critical role in investment decision-making, market regulation, and economic planning. This study proposes a hybrid prediction model that integrates Genetic Algorithm (GA), Whale Optimization Algorithm (WOA), and Long Short-Term Memory (LSTM) neural networks, referred to as the GA-WOA-LSTM model. In this framework, GA is employed to generate the initial population and perform global search for LSTM hyperparameter optimization, while WOA is applied to conduct local refinement of the search space. The LSTM model, known for its superior ability to capture nonlinear dependencies and long-term patterns in time series, is used to model and forecast future stock closing prices. The performance of the proposed model is evaluated on both training and test datasets using key metrics including Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), Root Mean Squared Error (RMSE), and the coefficient of determination (R2). Experimental results demonstrate that the GA-WOA-LSTM model significantly outperforms traditional baseline models in terms of predictive accuracy and generalization capability. This research offers a robust and effective modeling strategy for financial time series forecasting and provides valuable insights for real-world financial applications.