Kim Miskiw (SGEM) and Julius Grams presented their talk "The Potential of Deep Reinforcement Learning in Evaluating Market Design Options: Examples from the Power Market" at the Dialogtag Wirtschaft und Technik, the annual event of the KIT Department of Economics and Management, on 25 June 2026.

The core question: how can different power market designs be assessed before they are implemented in the real world?

Conventional simulations usually prescribe how market participants behave. With Deep Reinforcement Learning, agents learn their bidding strategies themselves and adapt to the incentives they face. This makes it possible to observe how changes to market rules affect strategic bidding, price formation, and market power. The approach is central to our current work in the ADAPT project and to the ASSUME toolbox (https://assume-project.de/).

The talk was part of a session on market simulation, which also covered spatial logistics networks and strategic information sharing in pathogen research. The discussion brought pointed questions from the audience.
Thanks to everyone who attended!