FedCSIS 2025 Challenge was entitled: Predicting Chess Puzzle Difficulty - Second Edition. It was the 11th data science challenge, organized within the scope of FedCSIS conference series. The goal was to build a model to predict the difficulty (measured as Lichess rating) of given chess puzzles. The 2025 competition was organized by:
- Ślęzak, Dominik, QED Software and University of Warsaw, Poland
- Ślęzak, Michał, Polish-Japanese Academy of Information Technology, Poland
- Świechowski, Maciej, Grail Team, Poland
- Zyśko, Jan, University of Warsaw, Poland
This year, 42 teams participated in the competition. They, collectively, submitted 1185 solutions. Team members represented 16 different countries from around the world, with largest number of representatives coming from Poland, Germany, United States and Singapore. After evaluation, the following contributions, found also in these proceedings, discuss the winning solutions:
- First place: Sebastian Björkqvist, Estimating the Difficulty of Chess Puzzles by Combining Fine-Tuned Maia-2 with Hand-Crafted and Engine Features
- Second place: Tyler Woodruff, Luke Imbing, Marco Cognetta, The bread emoji Team’s Submission to the 2025 FedCSIS Predicting Chess Puzzle Difficulty Challenge
- Third place: Szymon Miłosz, Pretraining Transformers for Chess Puzzle Difficulty Prediction
Top 3 teams received a FedCSIS 2025 registration and a cash prize: 1000$ for first place, 500$ for second, 250$ for third. Winning teams also had an opportunity to write an invited paper describing their solution, presented on the special session at FedCSIS.
