A hybrid price prediction method for carbon trading with multi-data fusion and multi-frequency analysis

Abstract

ABSTRACT Accurate price prediction for carbon trading is essential to provide the guidance for investment and production. The current prediction methods mainly depend on the carbon price itself, from which the change pattern of carbon price is studied. However, fusing the multi-source data, e.g. trading message and public sentiment, and taking proper data processing to improve the prediction accuracy need in-depth research. In this paper, a hybrid price prediction method utilizing both the statistical and intelligent models is established on multi-source data, and the data characteristics are fully explored by correlation analysis and multi-frequency analysis. The study on Guangdong market show that: the accuracy of proposed method is superior to the benchmark ones: root mean square error and mean absolute percentage error are reduced by 19.27% and 7.16%, while determination coefficient and trading return are increased by 8.31% and 25.11%. The proposed method is helpful for stakeholders to manage their trading. Graphical abstract