Interval Forecasting Approach And Application Research Of Carbon Price

Carbon markets have been increasingly applied in emission reduction practices as a cost-effective means of addressing climate change,including in China.However,the price volatility and frequency in carbon markets have adversely affected their effectiveness and emission reduction performance,and brought about many risks.The main focus of this paper is on interval forecasting analysis of carbon market prices in China,aiming to improve the accuracy of carbon market price predictions.The main innovative contributions of this paper include:(1)A residual-based bootstrap(RBB)and EMD-DEXGB-based interval forecasting model was constructed for univariate time series of carbon market prices.Firstly,the empirical mode decomposition(EMD)method was used to decompose the original carbon price series into simple mode sequences that are stationary,structurally simple,and highly regular.Secondly,extreme gradient boosting machine(DEXGB)optimized by differential evolution algorithm was used to predict each simple mode,and the corresponding distribution prediction was obtained through RBB.Finally,the prediction results of each mode were integrated to obtain the distribution prediction of the original carbon price,and the interval prediction result was obtained.The results show that this model can effectively mine the complex price fluctuation rules of carbon prices and obtain accurate prediction results,and can serve as a competitive and feasible solution for carbon price interval prediction.(2)An interval forecasting model based on BEMD-DEXGB was constructed for interval time series of carbon market prices.Firstly,binary empirical mode decomposition(BEMD)applicable to interval time series was introduced to decompose the complex carbon market price intervals into several simple components.Secondly,extreme gradient boosting machine(XGB)was used to predict the simple components after decomposition,and differential evolution algorithm(DE)was introduced to optimize all model parameters simultaneously.Finally,the model linearly integrated the prediction results of each component to obtain the interval prediction of carbon market prices.The empirical results show that,compared with the current popular prediction models,the proposed carbon market price interval prediction model in this paper can achieve better interval coverage rate and smaller prediction errors.(3)A trading strategy based on carbon market price interval prediction results was constructed.In this paper,a trading system was built,and the effective integration of the system with the output results of the prediction model was achieved.By directly using interval prediction results as input variables for the trading system,the performance of the trading strategy was evaluated,and the predictive ability of the model was further tested.This helps investors make clear investment decisions on trading timing and price selection.