Elo Rating Calculator

Frequently Asked Questions

How is the expected score in Elo calculated?

Expected score for player A = 1 ÷ (1 + 10^((Rb − Ra) ÷ 400)). A 200-point rating gap gives the favorite about a 0.76 expected score (76% in win-probability terms).

How is the new rating computed?

New Ra = Ra + K × (actual − expected), where actual is 1 win / 0.5 draw / 0 loss. Beating an equal opponent with K = 32 gains 16 points.

What does the K-factor do?

K controls volatility: a high K (32–40) moves ratings quickly (newer/provisional players), a low K (10–16) stabilizes established players. Chess federations lower K as rating and games played increase.

Why do points won equal points lost?

Elo is zero-sum: the winner gains exactly what the loser drops in a head-to-head, so the rating pool stays balanced and ratings reflect relative, not absolute, strength.

Provided by AllCalculators.io
Free online calculators for everyday. No registration required.

Informational only. Not betting advice.

Sports & Betting Disclaimer: Informational only. Not betting advice.

This calculator is for informational and educational purposes only and is not betting, gambling, or financial advice. It does not predict outcomes or guarantee returns. Gambling involves risk; never wager more than you can afford to lose. You are responsible for complying with the gambling laws in your jurisdiction.

JavaScript is required to use the interactive calculator above. The questions and answers below remain readable without JavaScript.

How This Calculator Works

This tool is for informational and educational purposes only and is not betting advice - please play responsibly. The Elo rating system, created by Arpad Elo for chess and now used across esports, soccer, tennis models, and countless online ladders, expresses each competitor's skill as a single number. The core idea: the rating gap between two players predicts the expected result, and after a game each rating moves toward the actual result by an amount that depends on how surprising it was. Beating a much stronger opponent gains many points; beating a much weaker one gains few; losing to someone you were expected to beat costs you heavily. Over many games these self−correcting nudges converge so that a player's rating settles near their true strength, and the difference between any two ratings becomes a calibrated forecast of how a match between them is likely to go. This calculator computes both players' expected scores and their updated ratings from a single game outcome, which is the atomic building block every larger Elo−style system - FIDE chess, FiveThirtyEight sports models, matchmaking queues - repeats game after game.

Formula / Method

Expected score for A = 1 ÷ (1 + 10^{(R_B − R_A) ÷ 400}), and E_B = 1 − E_A. The actual score S is 1 for a win, 0.5 for a draw, 0 for a loss. The rating update is R′ = R + K × (S − E), where K (the K−factor) controls volatility - larger K means faster, choppier adjustment. A 400−point gap corresponds to roughly a 10−to−1 expected result. Because the two players' score changes are equal and opposite when K is shared, the system is zero−sum: points won by one are lost by the other.

Worked Example

Player A is rated 1600, Player B 1500, and A wins (S_A = 1) with K = 32. Expected score for A = 1 ÷ (1 + 10^{(1500 − 1600) ÷ 400}) = 1 ÷ (1 + 10^−0.25) = 1 ÷ (1 + 0.562) ≈ 0.640. A was expected to score 0.64 and actually scored 1, so A gains 32 × (1 − 0.640) ≈ +11.5, new rating ≈ 1612. B was expected 0.360 and scored 0, losing 32 × (0 − 0.360) ≈ −11.5 to about 1488. If the lower−rated B had instead won, B would gain 32 × (1 − 0.360) ≈ +20 - a bigger swing because the upset was less expected.

Common Mistakes

Swapping the rating difference sign - the exponent uses opponent minus self for the expected score.

Using a draw score of 0 or 1 instead of 0.5, which corrupts both updated ratings.

Applying an oversized K−factor, making ratings whip around on single results.

Forgetting Elo is zero−sum with a shared K: one player's gain is the other's loss.

Treating an Elo win probability as a betting edge - it is a model estimate, not a market price.

Tips & FAQ

What K should I use? Chess uses 10–40 depending on level and experience; FIDE applies K = 40 for new players, 20 for most, and 10 for elite players above 2400, so ratings stabilize as a player's sample grows. Lower K for established players, higher for new or volatile ones. Why 400? It is the scaling constant Elo chose so a 400−point gap means about a 91% expected score and an 800−point gap about 99% - conventional, not fundamental; some systems use 480 or other bases. Does Elo predict the future? Only probabilistically; upsets are expected at exactly the rate the formula implies, and the rating only tracks demonstrated results, not potential or recent form beyond what games have shown. What if players have very different game counts? A provisional rating from few games is noisy; treat early numbers with caution and let K do its job of converging them. Use the expected score as a calibrated estimate, not a guarantee - any single match is uncertain, variance is large over short stretches, and if betting is involved, markets carry a persistent house edge that a model probability does not overcome. Set limits in advance, never chase losses, and seek help if it stops being entertainment.

Frequently Asked Questions

Related Sports & Betting Calculators