Gold Price Prediction Using Two-layer Decomposition and XGboost Optimized by the Whale Optimization Algorithm

Gold price prediction is of great importance in big data computing and economic sphere. This paper aims to contribute to the study of hybrid models that can be used to forecast the price of gold. In this study, The Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) is employed to decompose a residual term containing complex information following the variational modal decomposition (VMD) and an extreme gradient boosting tree (XGBoost) optimized by the Whale Optimization Algorithm (WOA) is combined to construct the VMD-RES.-CEEMDAN-WOA-XGBoost model. The closing price data of COMEX gold futures from 1 October 2018 to 20 November 2023 were selected as examples of gold futures price. A variety of factors that can affect the price of gold are considered in the research. This study indicates that the combined forecasting model proposed in this paper has superior performance when compared to the other comparison forecasting models evaluated. Furthermore, it has been found through SHAP analysis that the Nasdaq index, silver price, and the yield of US 10-year Treasury bonds are most closely related to the prediction of gold price.