Research On International Carbon Market Price Forecasting Based On Multiple Factors

As an indispensable commodity,the price change of carbon emissions will affect the production and operation of enterprises and the stable development of carbon trading market.The forecasting of carbon market price is a key issue of risk management for carbon financial market participants.Therefore,it is of great theoretical and practical significance to study the price forecasting of international carbon market.Due to the internatio nal carbon market price presents nonlinear,non-stationary and multi frequency irregular features,the traditional single model is difficult to fully characterize the carbon price fluctuation characteristics,multi frequency combination forecasting model can dig a variety of internal rules of the carbon price at different frequencies,thus better grasping the regularity of carbon price fluctuation.The optimal variational modal decomposition method(OVMD)is used to decompose EUA spot price into multiple mode components.Run-length-judgment method is used to reconstruct the decomposed series into low,medium and high frequency,the frequencies contains more focused information about carbon price.Aimed at the carbon price of each frequency sequence influenced by different factors and presents different fluctuation characteristics,existing research about carbon frequencies price forecasting,only consider their own history data,not completely cover all of the carbon price forecast information,may affect the carbon price forecasting accuracy.In existing research,the traditional BP neural network or support vector machine is used to forecast each carbon frequency price,it may lead to the problem of instability algorithm or running the program slow.Considering the KELM method has the advantages of stable prediction performance,high computational efficiency,KELM with different kernel functions are first introduced in the forecasting of the reconstructions of the carbon price.Lastly,add them together as the carbon price final forecasting value.The result shows that:(1)the introduction of factors into multi frequency combination prediction model is better than that only considering the carbon price of each frequency time series model,show that the introduction of C ER spot price,EUA futures price and coal price in the low carbon price forecasting,the introduction of CER spot price factors into the carbon median frequency price forecasting,can provide more useful information for the carbon price prediction,enhance forecasting accuracy of carbon price,the multi frequency combination model introducing factors is more suitable for the carbon market's actual situation;(2)corresponding KELM is introduced to forecast carbon price of each frequency series,can make full use of the advantages of kernel functions,a good solution to the problems of forecasting performance instability by using BP or computing speed needs to be improved by using SVM in existing research about carbon price forecasting;(3)decomposition of carbon price by using OVMD solves the mode decomposition of EMD carbon price aliasing problems effectively,making the carbon price change rules and hierarchical characteristics more clear;(4)The multi frequency combination model has good forecasting effect,which solves the problem that single model is difficult to describe the fluctuation of carbon price.The contribution of this paper is to develop theoretical research on the price forecasting method of international carbon market.It can provide useful theoretical reference for investors to make reasonable international carbon market decisions and avoid carbon market risk.