Carbon price prediction model based on multi-agent and environment co-evolution

Predictive models for carbon prices are of great significance for the development of a low-carbon economy. Many scholars have conducted carbon price prediction studies using various methods, but most of these studies have focused on the macro level without analyzing the micro level. The multi-agent and environment co-evolution model constructed in this paper uses neural networks to simulate enterprise decision-making. It forms market equilibrium prices based on the auction mechanism and adjusts the decision models of agents in a coordinated evolution manner based on these equilibrium prices, resulting in the final prediction outcomes. Additionally, this model can also be used to predict the prices of various futures, stocks, or other commodities, analyzing the preference changes of different agents from a micro-level perspective. This paper demonstrates through a case study that the multi-agent and environment co-evolution model with strong heterogeneity achieves higher accuracy in specific prediction tasks. • Integration of Multiple Methods : It combines multi-agent models, neural network models, genetic algorithms, and evolutionary game theory, providing a new perspective for carbon market price prediction. • Co-evolution Mechanism : It simulates the decision changes and interactions of market participants through the co-evolution of multiple agents and the environment. • Micro-level Analysis : By analyzing market behavior at a micro-level, it captures the complex dynamic behavior of the market, thereby improving the accuracy and stability of carbon market price predictions.