Enhancing Mutual Fund Price Prediction: A Hybrid Ensemble Approach with Random Forest, SVR, Ridge, and Gradient Boosting Regressors

This research paper introduces an innovative methodology for predicting mutual fund prices in the Indian financial market by utilizing a hybrid ensemble learning technique based on the stacking regressor algorithm. Conventional forecasting techniques frequently face difficulties capturing the intricate nonlinear connections and interdependencies within financial data. To tackle this problem, a suggested solution is the introduction of an ensemble learning framework that harnesses the collective capabilities of multiple base learners to enhance prediction accuracy. The proposed hybrid ensemble approach compares well with traditional single-model techniques and other ensemble methods. The ridge regressor is used as a meta-regressor in this proposed stacking regressor-based hybrid ensemble model. Our comprehensive evaluation reveals remarkable performance metrics, including a low mean squared error (MSE) of 0.0000169477447, root mean squared error (RMSE) of 0.0041167638687, mean absolute error (MAE) of 0.0025804022730, an impressive R-squared (R2) score of 0.9999829824919, and an explained variance score (EVS) of 0.9999832176376. The findings of this study are significant as they demonstrate that the stacking regression-based hybrid ensemble approach outperforms standalone models. It excels in terms of predictive precision, resilience, and consistency, thereby showcasing its potential for mutual fund price forecasting. Using ensemble learning techniques, we can improve the prediction accuracy of mutual fund prices. The proposed model shows promising results against individual prediction models and can be used for portfolio optimization, future price predictions and reduces the risk associated with mutual fund portfolios.