Elo Expected Score Calculator
Your win likelihood
76.0%
Opponent win likelihood
24.0%
How it works
The Elo rating system, developed by physicist Arpad Elo for chess, is used to calculate relative skill levels in competitive games. The core concept: each player has a numerical rating, and the system predicts the expected probability of winning based on the rating difference. After each game, ratings are updated based on whether the actual outcome matched the prediction. The Elo Expected Score Calculator computes win probability and post-game rating changes.
**Expected score formula** For a match between Player A (rating Ra) and Player B (rating Rb): Expected score for A: Ea = 1 / (1 + 10^((Rb − Ra) / 400)) Expected score for B: Eb = 1 − Ea
If Ra = 1500 and Rb = 1700: Ea = 1 / (1 + 10^(200/400)) = 1 / (1 + 10^0.5) = 1 / (1 + 3.162) = 0.240 (24% win probability for Player A).
**K-factor and rating update** New rating = Old rating + K × (Actual score − Expected score). Score: Win = 1, Draw = 0.5, Loss = 0. K-factor controls update magnitude: K=32 (beginners, larger swings), K=16 (established players), K=10 (masters). Example: Player A (1500) beats Player B (1700) — a 76% upset probability: New Ra = 1500 + 32 × (1 − 0.240) = 1500 + 24.3 = 1524.
**Elo systems in games** Chess (FIDE): standard Elo. Chess.com/Lichess: similar. League of Legends LP system: tier-based Elo variant. CS2 Premier: visible Elo rating (2023+). Overwatch 2: skill rating uses Elo. Rocket League: MMR (matchmaking rating) = Elo variant. The calculator works for any two-player Elo implementation.
Privacy: all calculations run in the browser. No account data is transmitted.
Frequently Asked Questions
- The divisor 400 in 10^((Rb-Ra)/400) was chosen empirically by Arpad Elo to calibrate the system: a player with a 400-point rating advantage wins approximately 91% of games (10^(400/400) = 10; expected score = 1/(1+10) ≈ 0.091 for the weaker player). A 200-point advantage: expected score ≈ 76% for the stronger player. The number 400 reflects the spread of chess skill ratings at the time Elo was calibrated — it's a scale constant, not a fundamental mathematical value. Some implementations (like Glicko) use 173.7 as the divisor to align with a standard normal distribution scale.
- Team games adapt Elo by treating each team as a single 'player' with a combined rating (average of team member ratings). The win/loss updates each member's individual rating by the same K × (actual − expected) formula. The challenge: individual performance within a losing team varies enormously. A team that loses 10 rated games in a row may have one player carrying while others underperform — pure win/loss Elo doesn't distinguish. Modern matchmaking systems (like TrueSkill, used by Xbox/Halo; Glicko-2, used by many games) add performance-based uncertainty modelling to address this.
- This typically means you're winning and losing at approximately the rate predicted by the system — you're correctly rated for your current performance level. A stuck rating isn't failure; it's calibration. Ways to increase: improve your win rate above the expected threshold. The K-factor determines how quickly your rating responds — beginners with K=32 move faster (32 points per decisive result against an equal opponent); established players with K=10 move more slowly. Some systems implement K-factor reductions as players become more established, making it harder to swing rating significantly after many games.
- FIDE (the international chess federation) uses Elo for standard chess ratings. However, FIDE made significant changes from the original Elo system: K-factor changes by rating level (K=40 for new players below 2300, K=20 for players between 2300–2400, K=10 for players above 2400). Additional bonus points for large wins. Floor ratings. FIDE ratings update monthly. Online chess platforms (Chess.com, Lichess) each maintain their own rating systems that differ from official FIDE — Lichess uses Glicko-2, which adds a rating deviation (uncertainty) parameter. A 2000 FIDE rating ≠ 2000 on Chess.com ≠ 2000 on Lichess.