基于Bagging集成算法的COMEX黄金期货价格预测

Prediction of COMEX Gold Futures Prices Based on the Bagging Integrated Algorithm

摘要: 针对全球金融市场复杂性问题，构建准确的黄金价格预测模型显得尤为重要。文章选取多个影响黄金价格的主要因素，考虑宏观经济、金融市场、政治风险等多个维度，运用斯皮尔曼分析法筛选指标并建立指标体系框架。采用灰狼优化算法(Grey Wolf Optimizer, GWO)优化极端梯度提升模型(eXtreme Gradient Boosting, XGBoost)，构建GWO-XGBoost模型，并将其作为基础学习器，建立集成模型Bagging-GWO-XGBoost。实验结果显示，优化算法提升了模型在COMEX黄金期货价格预测中的表现；当加入集成算法后，模型的RMSE、MAE和MAPE均有所改善。因此，该模型能够为黄金价格预测提供更加精准的预测数据，被视为一种实用高效的金融工具，帮助金融市场进行预测与决策。

Abstract: Given the complexities of global financial markets, constructing an accurate model for gold price prediction is of significant importance. This study selects multiple key factors influencing gold prices, considering various dimensions such as macroeconomics, financial markets, and political risks. Spearman’s correlation analysis is employed to filter indicators and establish an indicator system framework. The Grey Wolf Optimizer (GWO) is utilized to optimize the eXtreme Gradient Boosting model (XGBoost), resulting in the development of a GWO-XGBoost model. This model is then used as a base learner to construct an ensemble model, Bagging-GWO-XGBoost. Experimental results demonstrate that the optimization algorithm enhances the model’s performance in predicting COMEX gold prices. Furthermore, the integration of the ensemble algorithm leads to improvements in the model’s RMSE, MAE, and MAPE metrics. Therefore, this model can provide more accurate predictive data for gold price forecasting and is considered a practical and efficient financial tool, aiding in market predictions and decision-making processes.