基于滚动时间窗的碳市场价格分解集成预测研究

提高碳市场价格预测准确性对于交易风险监测以及碳市场平稳发展具有重要价值。针对复杂的、非线性碳市场价格数据的短期预测误差偏大、分解过程易产生数据泄露问题，提出了基于滚动时间窗的SSA-SVR分解集成预测框架。首先，选取时间窗数据，继而借助奇异谱分析将时间窗内碳价序列分解重构为高、低频序列；然后，使用支持向量回归方法对高、低频序列分别进行预测；最后，加和集成预测结果，得到下一时刻的碳市场价格预测值。通过不断更新时间窗的数据内容，动态执行“分解-预测-集成”过程，实现碳市场价格的实时预测。研究结果表明，本文所提出框架表现出优异且稳定的预测性能，在碳市场价格预测研究中具有良好的适用性和有效性。

A Decomposition Ensemble Model with Sliding Time Window for Forecasting Carbon Market Prices

Improving the accuracy of carbon market price forecasting is of significance for the monitoring of trading risk and the stable development of the carbon market. Aiming at the problems of large errors in the short-term forecasting of complex and nonlinear carbon market prices and data leakage in the decomposition process, a SSA-SVR decomposition ensemble prediction framework with sliding time window is proposed. Firstly, the time window data are selected, decomposed and reconstructed into high and low frequency sequences by using singular spectrum analysis and singular entropy. Then, support vector regression algorithm is used to forecast the high and low frequency sequences. Finally, the one step ahead of carbon market price forecasting value is obtained by adding and integrating the above results. By continuously updating the data content of the time window and dynamically executing the process of “decomposition-forecasting-integration”, real-time forecasting of carbon market price is realized. The empirical results show that the forecasting framework proposed in this paper exhibits satisfactory and stable forecasting performance, which is a suitable and effective tool for forecasting carbon market prices.