Chess Elo calculator
Enter two ratings and a K-factor to see what a win, a draw, and a loss are each worth. Useful before a tournament game, and useful afterwards for understanding why the number moved the way it did.
The game
Expected score: 36.0% — the share of the points the rating system predicts you would take against this opponent over many games.
What each result is worth
ResultChangeNew rating
- Win+12.81513
- Draw+2.81503
- Loss-7.21493
Each row is the rating you would carry away from this one game.
What the expected score means
Elo does not score you against an absolute standard. It scores you against a prediction. The expected score is what the rating system thinks you will average against this opponent over many games — 0.5 means an even match, 0.75 means it expects you to take three points in four.
Your rating only moves by the amount you beat or miss that prediction, scaled by the K-factor. This is why a 200-point favorite gains almost nothing for winning and bleeds rating for drawing: the system had already priced the win in.
Why chasing rating points is bad strategy
Once you can see the maths, the temptation is to farm it — play weaker opponents, avoid stronger ones, protect a number. It works briefly and then stops, because rating is a measurement of strength rather than a cause of it. A rating built by avoiding hard games reverts as soon as you stop avoiding them.
The more useful reading is the reverse. Losses to much stronger opponents cost you very little, which makes them close to free information about what you cannot yet do. The expensive results are the ones against opponents at your own level, and those are decided by the mistakes you make repeatedly, not by the pairing.
Elo, Glicko, and what your site uses
Arpad Elo’s system assumes a fixed K per player. Modern online pools mostly use Glicko or Glicko-2, which add a rating deviation: a number representing how confident the system is about you. New or inactive accounts have a high deviation and move in bigger steps until the system settles on them.
That is why your online rating can jump further than this calculator suggests, especially in your first games or after a break. The underlying logic — you gain by beating expectation and lose by missing it — is identical.
Frequently asked questions
Your expected score against an opponent is 1 divided by (1 plus 10 to the power of the rating difference over 400). Your rating change is the K-factor multiplied by the difference between what you actually scored and that expected score.
It depends on the pool. FIDE uses 40 for new players with fewer than 30 games, 20 for most rated players, and 10 for players who have reached 2400. Online sites set their own, and often vary it by time control, so treat the result here as an estimate unless you know your pool.
Because the system already expected you to win. If your expected score is 0.9 and you win, you only banked 0.1 more than predicted. The same asymmetry is why a single loss to a much weaker opponent costs a lot.
Not exactly. Chess.com uses a Glicko variant and Lichess uses Glicko-2, both of which track a rating deviation alongside the rating so uncertain ratings move faster. The direction and rough size of the change still follow the same logic this calculator shows.
Provisional ratings, rating floors, and deviation-based systems all change the size of the step. Use this to understand the shape of the math — favorites gain little, upsets move a lot — rather than to predict an exact number on a specific site.
The rating is the scoreboard, not the diagnosis. If you want to know which recurring mistake is actually holding the number down, that takes a look at your own games.
Analyze my last 20 gamesOther free tools
Analysis board
Play through any position with Stockfish running in your browser. Paste a FEN or PGN, branch off, and see the evaluation move by move.
Chess clock
Time an over-the-board game: blitz through FIDE classical presets, increment or delay, the 40-move split, and a screen it asks to stay awake.
Board visualization
Train the board in your head: square colors, coordinates both ways, knight routes, a short opening line followed blindfold, and a real position held from memory — as a set or a 30-second sprint.