Blog
Long think, wrong think: Olympiad edition
Long thinking time in the 2026 Chess Olympiad often preceded mistakes.
Rating deflation: today's chess elite are underrated
Have classical ratings deflated, such that a given Elo rating now corresponds to stronger play? According to XPA, the answer is clearly yes.
Rules for annotations at ChessDS
Describing the ten rules that produce annotation symbols at ChessDS.
Time control and playing strength in St. Louis
What's the difference between classical and blitz for top players, in classical Elo? Seeking an answer in St. Louis.
Testing XPA: Elo ratings, accuracy, and a hint of deflation
Comparing XPA data with Elo ratings and accuracy %. XPA tracks Elo better than existing measures, and it seems to show evidence of rating deflation since 2005.
Measuring chess performance with ChessDS XPA
A new metric for player performance, built on result and move predictions instead of raw engine evals.
Predicting chess with machine learning
A blog post describing the two production models: a calibrated win/draw/loss predictor and a human-move ranker for 2400+ classical chess.
