Estimated chess rating: how your performance is calculated

How ChessAnalyzer estimates a rating from the quality of your moves, why it shows a range, and how it differs from official ratings.

The Insights screen shows an estimated rating calculated from the quality of your moves in the games you analysed. It is a useful mirror, but it is an estimate, and this page explains how it is built and what its limits are.

Performance per game

For each analysed game, ChessAnalyzer estimates a performance from your accuracy and how many serious errors you made. A clean game with no mistakes produces a high performance; a game decided by one blunder produces a lower one, even if the rest was excellent.

Combining many games

Games are not all equally informative. Longer games and deeper analyses count more; very short games (under eight moves) are ignored. The estimate is also pulled gently towards the ratings recorded in your games' PGN headers, so a single unusual game cannot swing it wildly.

Why there is a range

With few games, any estimate is uncertain. ChessAnalyzer shows a range around the estimate — roughly two standard errors — and a confidence level. The range narrows as you analyse more games. Treat a "low confidence" estimate as a first impression.

What it is not

The estimate is not a FIDE, national or online-site rating. It measures move quality as judged by an engine, not results. Players who win by setting practical problems for their opponents can perform above their accuracy, and vice versa.

Making it useful

Analyse every serious game, not only the ones you lose.
Use the same analysis mode so games are comparable.
Watch the trend over weeks, not the number after one game.