Build an AI Training Agent to play the Pokémon Trading Card Game. This project aims to enhance the performance of an AI Training Agent with the Pokémon Trading Card Game (TCG). The research focuses on training AI Training Agents for competitive play in a system where probability, unknown elements, and strategic planning are key determinants of success.
Note: This TCG AI Battle Challenge has two competitions. This competition is the Simulation competition. Learn more about the Hackathon here. Participation in the Hackathon is not required to enter this competition. As a strategic game, Pokémon TCG players must make gameplay decisions while being mindful of their opponent's own strategies, decks and hands. Various Pokémon types and thousands of different card combinations introduce gameplay variables, alongside card draws and coin tosses. The AI Training Agent’s training will be conducted within this competitive simulation framework. Not knowing what cards an opponent holds presents a core challenge. Participants are encouraged to explore novel methodologies for strategy learning and decision making.
Participants will be provided with a simulator (SDK) for training and testing. Using rule-based programming alone may not ensure a high ranking. Winning requires forward thinking, real-time adaptation, and optimal decision-making. The AI Training Agent must demonstrate high analytical capacity, adaptability, and be ready for the unexpected.
Evaluation Each day your team is able to submit up to 5 agents. Each submission will play Episodes against other agents on the ladder that have a similar skill rating. Over time skill ratings will go up with wins or down with losses. We only track the latest 2 submissions and use those for final submissions.
Timeline - June 16, 2026- Start Date. - August 9, 2026- Entry Deadline. - August 16, 2026- Final Submission Deadline. - August 17, 2026 to (approx.) August 31, 2026- Final Evaluation period.
Prizes The Competition track itself does not include monetary prizes. However, participants who submit a report to the Hackathon track will be eligible for prize awards. Final rankings for Hackathon prizes will be determined based on both the Competition leaderboard performance and the Hackathon evaluation.
How to Submit Submissions need to be a .tar.gz bundle with main.py at the top level directory and include a deck.csv. Upload this under the My Submissions tab.