ServedBets

Free tool

Elo Win Probability Calculator

Enter two Elo ratings (from our own Elo Rankings or anywhere else) and see the win probability the gap implies.

Rating gap
-
Player A win prob.
-
Player B win prob.
-
Embed this calculator on your own site

Free to use anywhere. A small "Powered by ServedBets" link stays on the embed. Paste this where you want it to appear:

How this is calculated

The standard Elo win-probability formula on a 400-point scale, the same one our own surface-specific Elo model uses:

P(A wins) = 1 / (1 + 10^(−(EloA − EloB) / 400))

A 100-point gap works out to roughly 64%, 200 points to roughly 76%, and 300 points to roughly 85% for the higher-rated player. This calculator doesn't know anything about tennis specifically. It's the general Elo formula, applied to whatever two ratings you enter.

Pull real, current numbers from the Elo Rankings page, which is filterable by surface. Remember to compare hard-court Elo to hard-court Elo, not overall to overall, for a hard-court match.

Worked example: 2000 vs 1850

  1. Rating gap: 2000 − 1850 = 150 points.
  2. P(A wins) = 1 ÷ (1 + 10^(−150 ÷ 400)) = 70.3 % in a best-of-three match.
  3. At a Grand Slam (best of five) the same gap is worth more, because the better player has more sets to turn it around: our model converts it to 74.8 %.
  4. Compare that with the market after removing the margin; a gap of a few points is normal noise, not automatically value.

Common mistakes

  • Mixing scales: Elo numbers from different sources are not comparable. Use both ratings from the same system.
  • Using overall Elo for a clay match. Surface ratings exist for exactly this reason.
  • Trusting ratings built from very few matches. Our model only uses its own rating after 15 tracked matches.

In tennis

Elo is one of the best freely available predictors for tennis matches, but it knows only results: not injuries, fatigue or motivation. We measured, for example, that players coming back from a retirement win clearly less often than their rating says.

Read next: ATP & WTA Elo rankings · Why ratings must be surface-specific · Head-to-head records

See today's picks live

Every model output, tracked and discussed in real time. Join the 2,647+ member community where it all happens.

Join Discord (opens in a new tab)