<![CDATA[Forecasting stock price movement is a complex task due to nonlinear patterns, market volatility, and the influence of various technical and fundamental factors. This study proposes a hybrid forecasting framework that integrates the sequential learning capability of the Gated Recurrent Unit (GRU) with the nonlinear regression strength of Extreme Gradient Boosting (XGBoost) to predict the daily closing price of Bank Central Asia Tbk (BBCA). The dataset consists of historical BBCA prices from 2017 to 2025 and includes technical indicators such as moving averages, RSI, MACD, and Bollinger Bands. An 80:20 chronological split was used to evaluate model generalization through MAE, RMSE, MAPE, and R² metrics. Experimental results show that the hybrid GRU–XGBoost model outperforms both standalone GRU and XGBoost models, achieving the best performance with MAE of 229.09, RMSE of 312.26, and R² of 0.874 MAPE of 2.37%. Furthermore, SHAP-based explainability analysis highlights that price-based features and trend–momentum indicators contribute most significantly to the prediction output, while the GRU-derived sequential feature enhances temporal pattern recognition. These findings demonstrate that combining deep learning and boosting techniques produces a more accurate and interpretable forecasting model suitable for financial decision-making and risk analysis.]]>
Copyrights © 2026
