Pickleball ratings explained

A rating is one number that sums up how a player has done against other players. It comes from results, so it is only as sound as the results behind it, and it says nothing a scoreboard did not already say.

Two numbers from two systems are two different things. A 4.0 in one system is not a 4.0 in another, and this page gives no conversion between them.

A rating and a skill level are also different. USA Pickleball describes a skill level as based on skills you can show and a rating as built from match results. Both are covered below.

Elo is the method behind PickleCrew's rating. For one game it needs two things: how likely each side was to win, and what happened.

As a formula, with ratings in Elo points: expected = 1 / (1 + 10^((other average − own average) / 400)), and change = round(K × (result − expected)). The calculator runs exactly this.

The working, one player at a time:

Side A was expected to win 70% of the time (0.7034). Side A won, which scores 1. 1 minus 0.7034 is 0.2966. K is 20 for 35 rated games, and 20 times 0.2966 is 5.93, which rounds to 6. So Side A gains 6 and goes from 1,500 to 1,506.

Side B was expected to win 30% of the time (0.2966). Side B lost, which scores 0. 0 minus 0.2966 is −0.2966. K is 32 for 12 rated games, and 32 times −0.2966 is −9.49, which rounds to −9. So Side B loses 9 and goes from 1,350 to 1,341.

The working. Side A's average is (1,500 + 1,100) ÷ 2 = 1,300, and Side B's is (1,400 + 1,300) ÷ 2 = 1,350. Each player uses their own side's average against the other side's. A2 is the one to watch: with only 3 rated games, A2's K is 48, so the same surprise moves A2 more than A1.

A1 was expected to win 43% of the time (0.4285). A1 won, which scores 1. 1 minus 0.4285 is 0.5715. K is 20 for 35 rated games, and 20 times 0.5715 is 11.43, which rounds to 11. So A1 gains 11 and goes from 1,500 to 1,511.

A2 was expected to win 43% of the time (0.4285). A2 won, which scores 1. 1 minus 0.4285 is 0.5715. K is 48 for 3 rated games, and 48 times 0.5715 is 27.43, which rounds to 27. So A2 gains 27 and goes from 1,100 to 1,127.

B1 was expected to win 57% of the time (0.5715). B1 lost, which scores 0. 0 minus 0.5715 is −0.5715. K is 32 for 15 rated games, and 32 times −0.5715 is −18.29, which rounds to −18. So B1 loses 18 and goes from 1,400 to 1,382.

B2 was expected to win 57% of the time (0.5715). B2 lost, which scores 0. 0 minus 0.5715 is −0.5715. K is 20 for 40 rated games, and 20 times −0.5715 is −11.43, which rounds to −11. So B2 loses 11 and goes from 1,300 to 1,289.

At the far ends of the range the same formula gives plain numbers. If a 1,000 beats a 2,100, both with 30 or more rated games (a K of 20), the 1,000 gains 20 and the 2,100 loses 20. If the 2,100 wins, both change by 0, because the result was nearly certain and each change rounds to a whole number.

Each player's change is rounded on its own. When two players share the same K, one gains exactly what the other loses; a scan of every gap from 0 to 2,400 points, in half-point steps, at K of 20, 32 and 48 found no case where they differ. When the Ks differ, the game is not zero-sum: a new player with K 48 who beats an established player gains more than the established player loses.

PickleCrew's rating is an informal crew rating, not a sanctioned one. These are the rules the app follows, each pinned by a test against its code.

DUPR says a win can still lower a rating, because a rating moves with the result against what was expected, not with the win alone.

DUPR, as of its March 2025 post (published March 19, 2025; read November 2, 2026): a scale from 2.00 to 8.00, with ratings given to three decimals such as 7.218. The post says a rating moves with performance against the expected score, that club and tournament matches carry more weight than recreational ones, that more matches make a rating steadier and that recent matches count more. It names no formula.

DUPR, as of its August 2026 post (dated August 7, 2026; read November 2, 2026): every match has an expected score from the players' ratings, and a rating moves by how the result compares with it, not by the win alone. DUPR's own example is winning 11-8 when 11-3 was expected, which lowers the rating. It lists four factors: performance against expectation, type of match, recency and match volume. It gives no formula.

UTR-P is described by UTR Sports as a modified Elo rating on a scale from 1.00 to 10.00, and it counts points won per game.

UTR-P, as of UTR Sports' help-center page on the UTR-P algorithm (undated; read November 2, 2026): a match rating comes from the rating gap between opponents and the points won per game, and a player's rating is a weighted average of up to 60 recent matches from 2022 on. One match gives a projected rating, and about five matches make it fully reliable. The page gives no formula.

USA Pickleball names DUPR as the official exclusive rating system for USA Pickleball-owned events on its player ratings page, and says it also sanctions tournaments that use either DUPR or UTR ratings.

USA Pickleball, as of its player ratings page (page updated September 16, 2026; read November 2, 2026): the page states that DUPR is the official exclusive rating system used across USA Pickleball-owned events, and its FAQ adds that USA Pickleball also sanctions tournaments that use either DUPR or UTR ratings.

These are separate scales from separate owners. This guide gives no conversion between them, and the calculator does not compute a DUPR or a UTR-P rating.

USA Pickleball, as of its skill levels page (page updated September 28, 2026; read November 2, 2026): its levels are based on skills and strategy you can show, not on match results, and its descriptions are not official ratings. It describes levels from 1.0 to 5.5 and offers a 12-question self-assessment questionnaire.

If you are looking for your level, start there. PickleCrew's calculator will not tell you a level: it shows what one game does to a PickleCrew rating.

A rating helps in two places: choosing who plays together, and checking that a night was fair. The round robin chart, the mixer and Mexicano and Americano plan partners without any rating, so a gap in level shows up in the scores, not in the draw. King of the court moves winners up and losers down, so a night sorts itself by results.

To see what one game would do before it is logged, use the rating calculator. It saves nothing and sends nothing.

A PickleCrew rating changes when a member logs a finished game in a crew, or when a ranked competition game is scored. An unranked competition game does not count. Sign in or create an account to join a crew and log games.

The calculator never changes anyone's rating, and nothing you type into it is saved.