Forecasting stock market time series through the integration of bee colony optimizer and multivariate empirical mode decomposition with extreme gradient boosting regression

Abstract

Stock price prediction is essential for the optimization of investment strategies, the mitigation of risks, and the facilitation of informed decision-making. Accurate forecasting is exceedingly difficult due to the nonlinear, nonstationary, and volatile nature of stock prices. This complexity is frequently not adequately addressed by conventional methods, underscoring the necessity of sophisticated hybrid models. This study uses stock price data from the Standard & Poor's 500 Index to develop a novel hybrid model, Multivariate Empirical Mode Decomposition-Artificial Bee Colony-Extreme Gradient Boosting Regression. Extreme Gradient Boosting Regression captures intricate patterns in the data, Artificial Bee Colony optimizes the hyperparameters of Extreme Gradient Boosting Regression to enhance model robustness, and Multivariate Empirical Mode Decomposition decomposes complex financial time-series data into manageable intrinsic mode functions. Close price, Momentum, Simple Moving Average, Moving Average Convergence Divergence, Relative Strength Index, and Trading volume comprise the dataset. These features are indispensable for identifying both short-term fluctuations and long-term trends. The presented model is significantly more effective than traditional models, as evidenced by its test set coefficient of determination of 0.9914. The proposed model's robustness is confirmed by comprehensive 5-fold cross-validation and ablation studies, which also emphasize the significance of its integrated components. Furthermore, the model's adaptability is further illustrated by its ability to generalize to other markets, as evidenced by its coefficient of determination values exceeding 0.99 on three other indexes. These results underscore the potential of artificial intelligence-driven hybrid models to enhance stock price forecasting, offering useful insights for policymakers, financial analysts, and investors.