Dice Question Streamline Icon: https://streamlinehq.com

Endogenize the offset credit price within the RL market model

Endogenize the greenhouse gas offset credit spot price within the current Nash-DQN multi-agent market framework by modeling price formation jointly with agents’ trading and generation actions rather than prescribing an exogenous price process, and investigate learning-based approaches to price endogenization in model-free RL.

Information Square Streamline Icon: https://streamlinehq.com

Background

In the presented model, the OC price follows an exogenous Brownian bridge with price impact from generation. The authors identify endogenizing price formation as an open problem, noting limited existing literature on endogenization within model-free RL.

Achieving endogenous pricing would more closely capture market microstructure and feedback effects between agent behavior and prices, potentially advancing RL methods for financial market modeling.

References

Within the current framework, there remain open problems that are worthwhile investigating. Finally, while our model includes price impact, the OC price is given exogenously. Endogenizing the OC price within the current setting would be a very interesting avenue to explore.

Multi-Agent Reinforcement Learning for Greenhouse Gas Offset Credit Markets (2504.11258 - Welsh et al., 15 Apr 2025) in Section 6 (Conclusion)