A Study On The Prediction Of The Volatility Of The EU Carbon Emission Market

As the largest and most mature carbon emission trading market in the world,the development and improvement of EU carbon emission trading market is of great significance to other countries in the world.The price volatility of carbon emission market is very important to the market participants.The prediction of the price volatility of carbon emission market will help the market participants to form the expectation of the volatility of carbon emission rights,so as to slow down the market panic.In this paper,the decomposition reconstruction prediction method is used to make short-term prediction of the price volatility of the European Union carbon emission right futures market.At the same time,the volatility is decomposed in detail,and the relationship between the decomposed volatility component and the original volatility is analyzed to explore the impact of each volatility component on the overall volatility.In this paper,OHLC range volatility is used as the estimation of the volatility of carbon emission rights,EMD empirical mode decomposition method is used to decompose the volatility,BP neural network,GA Optimized BP neural network and SVM support vector machine method are used to predict the volatility component,and dynamic rolling prediction and dynamic non-rolling prediction are used to predict the volatility.By comparing various forecasting methods,this paper explores the best short-term forecasting method for the volatility of carbon emission right price in EU.The main conclusions of this paper are as follows: 1.After the decomposition of volatility,the contribution of volatility components to OHLC range volatility is measured by approximation,similarity and volatility contribution factors.The results show that short-term volatility has the strongest ability to describe OHLC range volatility,followed by long-term volatility,and medium-term volatility.2.According to the fitting results of the model,the decomposition of volatility is helpful to improve the prediction accuracy,and the reconstruction of the decomposed volatility component can improve the prediction accuracy further.3.Compared with BP dynamic rolling prediction method,GA Optimized BP dynamic rolling prediction method and GA Optimized BP dynamic non rolling prediction method,it is concluded that BP dynamic rolling prediction method has advantages in prediction accuracy and prediction time.4.Compared with SVM,BP neural network is more suitable to predict the data with complex nonlinear characteristics,and SVM is more suitable to predict the data of relatively stable trend.5.Removing the noise component of short-term volatility can improve the accuracy of short-term volatility prediction.6.The prediction results of OHLC show that the prediction value of OHLC range volatility obtained by decomposition reconstruction prediction method can reflect the average volatility trend of the true volatility,but the prediction of abrupting component is poor.After the decomposing and reconstructing process,the prediction based on data characteristics can improve the prediction accuracy of volatility.