A study on the carbon emission futures price prediction

Abstract

This study focus on predicting prices of carbon emission futures using a range of methodologies, including traditional ARIMA models and machine learning algorithms. Our analysis uses data from 2005 to 2023 and encompass variables such as the Dow Jones Industrial Average, GDP, crude oil and natural gas futures, per capita carbon emissions, and industrial indices. We find a significant correlation between economic indicators and carbon futures prices. GDP and the Dow Jones have the most influence. Further, we observe that Machine learning models, especially Random Forest Regressor and Gradient Boosting Regressor, outperform traditional ARIMA models in predicting carbon futures prices. This highlights the effectiveness of modern approaches in understanding complex market dynamics. Additionally, the feature importance analysis emphasizes the critical role of economic variables in predicting carbon futures prices. Overall, the study provides valuable insights into the carbon emission futures market and offers implications for stakeholders in managing environmental risks and promoting sustainable development.