Elo Rating Calculator
Expected score and new Elo ratings for two players or teams after a win, draw or loss, with the K-factor your federation or league uses.
Each side’s expected score from the rating difference, both new ratings and the change after the result, and the change for every possible result.
Example: A 1500 player beats a 1700 player with K = 32: expected score 0.240, so the winner gains 24.3 to 1524.3 and the loser drops to 1675.7.
Beat the
expectation.
How the rating difference becomes an expected score, how K turns the surprise into points, and what the system assumes.
Expected score
In the logistic form of Arpad Elo’s system, a player rated D points above an opponent is expected to score 1 ÷ (1 + 10^(−D/400)) of a point: 0.5 when level, about 0.76 at +200 and 0.91 at +400. Expected score is not a win probability — draws count half.
The update
After the game each rating moves by K × (actual score − expected score), with 1 for a win, ½ for a draw and 0 for a loss. An upset moves ratings a lot; a win over a much weaker opponent barely moves them. Both sides move by the same amount when they share a K.
K is a choice
K sets how fast ratings react. FIDE, the USCF, national leagues and sports models each choose their own, often by rating level or number of games played, so the page asks for it.
What the numbers do not settle
Federations add rules this page does not: rating floors, caps on the difference used, provisional periods and rounding. Inputs stay in your browser; the same anonymous usage counts as the rest of the site apply.
SOURCES
- Arpad E. Elo, The Rating of Chessplayers, Past and Present (1978)
Last reviewed 24 September 2026. How results are checked: How we verify.