Padel Elo Calculator — doubles ratings that show their work

Four steps, and all four constants are on this page. First, each side gets one number: a team rating is the mean of its two players, so 1100 and 1000 is a team of 1050. Second, the expected result comes out of the Elo curve — the chance Team A wins is 1 ÷ (1 + 10 ^ ((Rb − Ra) ÷ 400)). That 400 is Elo's scale constant, the thing that makes a 400-point gap a nine-in-ten favourite, and it is the same denominator chess federations have used since the 1960s.

Third, the margin. The games score is folded in as ln(|games difference| + 1) × (2.2 ÷ damp), where damp is (winner's team rating − loser's team rating) × 0.001 + 2.2, floored at 0.5. That multiplier is ours, not anybody's published system: it rewards a convincing scoreline and then damps the reward when the rating gap was already wide, so a strong pair thrashing a weak one is not paid twice for the same mismatch. The idea is borrowed from the margin-of-victory multipliers public sports Elo systems use; the shape and the numbers are our choice.

Fourth, the move: points = K × margin × (actual − expected), rounded to a whole number, where actual is 1 for the winning side and 0 for the losing one. Both players on a side move by that same amount, and the losing side moves by exactly the negative of it, so the four ratings always sum to what they summed to before. K is yours to pick — 32 for casual nights, 24 for an established crew, 16 for a long history.

Worked example, the one the page loads with ?demo=1: 1100 and 1000 (team 1050) against 950 and 1050 (team 1000), K = 24, first pair winning 6–3. Expected 57.15%, damp 2.25, margin 1.355, so 24 × 1.355 × (1 − 0.5715) = 14 points. 1114 and 1014 against 936 and 1036. The displayed win probability is clamped to 1–99% because nothing on a padel court is certain.