How the Umpiry Score & Rating Work

What it tells you

How likely each outcome is before the first point. The two values always add up to 100%.

What it does not mean

It is not a certainty, not a bookmaker price and not a live, in-match probability.

How to read it

68% means the model estimates this player wins about 68 times out of 100 in comparable situations. The model is calibrated: we test it on matches it has never seen, across the whole probability range.

What it uses

Pre-match information only: both players’ ratings, overall and on the surface, player profile factors, and the match context — surface, round, tournament level and tour. Betting odds are not an input.

One model per tour

Tours differ, so ATP and WTA tour level, men’s and women’s Challengers, and men’s and women’s ITF events each have their own model.

Updated

Every hour until the match starts. The last estimate before the first ball is frozen: wherever a live or finished match shows a probability, it is that pre-match number, never a hindsight revision.

Reading a probability

50%

A balanced matchup.

70%

A clear favourite, not a certainty. In 10 comparable matches the other player would be expected to win about 3.

30%

An underdog who can legitimately win. An underdog win is an expected outcome, not a model error.

What it tells you

How strong each player’s position in this matchup looks once the circumstances are taken into account. Each side is scored on its own.

Matchup

Longer-term standing, how each player performs on the surface being played, and their previous meetings. The backbone of the score.

Form

Recent results and momentum.

Fatigue

The physical toll of recent tennis: long matches, a heavy run of matches and deep tournament runs.

Travel

Crossing time zones, playing at home or how the player has fared in the host country, and a change of surface between tournaments.

Motivation

Context around the event itself: how far the player went here a year ago, a return from injury, and entering through qualifying or on a wildcard.

What it does not mean

It is not the chance of winning, the two values do not add up to 100, and it is not a betting tip.

How to read it

Compare the two players’ scores with each other, never with Win Probability. No single factor decides it; when part of the data is missing, the score says how many of its signal groups it was based on.

Updated

Every two hours before the match, from information available before the first ball. It is shown before and during the match and removed once the match is over.

How it works

We simulate the actual draw to get every player’s title chance, then run the same simulation over many random draws — the field shuffled into the bracket with no seeding — to get their chance in an average draw. The difference is the draw luck.

How to read it

Above zero means a kinder path than an average draw, including the advantage a top seed gets from being kept apart from the other contenders. Below zero means a tougher route.

What it does not mean

It is descriptive context about the bracket, not a prediction or a betting tip.

Updated

Fixed once the bracket is set: it is a property of the draw as it was made.

Overall rating

One rating from every eligible singles match at ATP, WTA, Challenger and ITF level. Team competitions, the Olympics, juniors, exhibitions and wheelchair events are left out.

Surface ratings

Separate hard, clay and grass ratings, calculated the same way from matches on that surface only. Indoor hard counts as hard; carpet moves only the overall rating.

How it moves

Everyone starts at 1500. After each match the winner takes points from the loser, more for beating a stronger player. A new player’s rating settles after roughly 30–50 matches.

Match weight

How far one result moves a rating depends on the event: Grand Slams and the Tour Finals 32, Masters 1000 and WTA 1000 24, other tour events 16, Challengers and ITF 8.

How to read it

Higher means stronger, and the gap between two players matters more than the number itself. Other Elo systems, the UTR and ranking points use their own methods, so values are not comparable across systems.

What it does not mean

It is not the official ATP or WTA ranking and does not replace it.

Updated

Every day from the previous day’s results, with a full recalculation each week.

Official ranking vs Umpiry Rating

The two answer different questions. Neither replaces the other.

Serve

How effective a player’s service games have been over the last 52 weeks at ATP and WTA level, adjusted for the strength of the opponents faced. Also rated per surface.

Return

The same on the opponent’s serve: how effective a player’s return games have been.

Clutch

Performance in the moments that decide matches — break points saved and converted, deciding sets and tiebreaks — at ATP, WTA and Challenger level since 2005, with recent seasons counting more.

How to read it

Higher is stronger, relative to the rest of the tour; 100 is the best current value. Compare players on the same rating, not across ratings. Clutch measures pressure performance, not overall strength.

Updated

Every day.

Minimum sample

Serve and Return

At least 1,000 service (or return) points in the last 52 weeks; 450 for a surface rating.

Clutch

At least 100 matches and 200 break points faced; each pressure situation also needs its own minimum.

Below the minimum

There is no rating: the player is left out of that list instead of being shown with 0.

Common Opponents

Each player’s results against the same opponents over the 12 months before the match, or before today on a head-to-head page. A better record is context, not proof that a player is better.

Freshness

A 0–100 snapshot of physical workload, where higher means more rested. It counts the games played in the last two weeks, with recent and long matches weighing more, and stays blank after three weeks without a match.

What it does not mean

Freshness measures recent load, not form or quality. It is about the player, not about how current the data is (that is data freshness, below).

Updated

Freshness every day; common opponents from the latest results.

Eligible matches

Completed singles matches at ATP and WTA tour level (Grand Slams and WTA 125 included), at Challenger level, and in the Davis Cup and Billie Jean King Cup. Qualifying counts; a retirement is a result. Walkovers, cancelled and suspended matches neither extend nor break a streak. ITF, junior and exhibition matches do not count.

Who appears

Players with an official ATP or WTA ranking of 250 or better who played an eligible match in the last 30 days. Results are read from the last 365 days.

Streaks

Consecutive wins (or, for a cold streak, losses) back from the latest eligible match. Streaks of three or more are listed, longest first; ties go to the more recent streak, then the better ranking.

Form movers

Over the last 10 eligible matches, each result is compared with what both players’ Umpiry Ratings expected, and the differences are averaged: Form +24 means the player won 24 percentage points more often than expected. At least 8 of the 10 need a known rating for both players, and +10 or more is listed.

Updated

Every hour. “All” merges the ATP and WTA lists in the same order; it is not a combined ranking.

What it tells you

Which player has had the stronger recent play. With point-by-point data it follows the most recent points; otherwise the last eight games, where a break of serve counts more than a hold.

What it uses

Points and games only. It uses no betting odds and no win probability.

States

Leaning to one player · Even · Too early. “Too early” is shown until there is enough play to read.

What it does not mean

Momentum is not Win Probability and does not predict the result. A single break of serve can turn it around in minutes.

Updated

During the match, point by point or game by game. After the match it shows how the momentum ran at the end.

What it tells you

The lower the winner’s pre-match chance, the higher the score: if the model gave the winner 22%, the result scores 78. 70 or more marks a genuine shock — the winner had 30% or less.

What it does not mean

It does not say how good, close or important the match was.

Coverage

From 5 June 2026, when the win probability model went live. Earlier matches have no frozen probability, and we do not assign one after the fact.

Updated

Once, when the match is completed, from the probability frozen at the start.

Surprise Score ≠ Match Greatness Score

A surprising result is not automatically a great match, and a great match is not automatically a surprise.

What it tells you

Which completed matches rate highest. A Grand Slam final that goes the distance between two of the best in the world scores highest; a routine early-round win scores low, and a typical tour-level match lands in the 20s.

What it does not mean

It is not Win Probability, not a percentage and not a percentile. The scale is fixed, so a match keeps its number whatever is played later; only a handful of matches since 2000 reach 100.

Eligibility

Completed main-draw singles matches at ATP and WTA tour level (Grand Slams, the Finals, 1000, 500 and 250 events) and at the Olympics, since 2000. A retirement is rated once two sets were completed; walkovers, incomplete score lines and qualifying matches are not.

Reasons on a card

Facts from the score line, the official ranks at the time of the match (available from 2025) and the stage of the event, such as “Five sets” or “Won from a set down”.

Match of the Day, Week and Year

The highest-rated match of the day (Central European time), of the last seven days and of the calendar year.

Updated

Every hour; a completed match appears on the page within about two hours. Live matches are not rated.

Rankings

Official rankings and Umpiry ratings are different systems and are always labelled.

Official rankings

ATP and WTA singles rankings as published by the tours, updated weekly. Live rankings are recalculated every hour.

Umpiry ratings

Umpiry Rating (overall and by surface), Serve, Return and Clutch.

Age and country lists

Built on the official rankings. Age lists keep the official tour rank; country lists number players within the country and show the tour rank beside it.

Rating lists

Umpiry Rating lists include players in the official top 500; Serve and Return lists the top 150.

Defending

The points dropping out of the player’s total over the next twelve weeks: the points earned at the same events a year ago, which expire as those events come round again.

Win …

The points the player would gain by winning the round they are playing now.

Expected

The most likely ranking once the current tournament finishes: we simulate the rest of the bracket with Umpiry Ratings and take the most probable outcome.

What it does not mean

All three describe what the published ranking formula does with results that already exist or with one named match. None predicts who wins a tournament, and none is advice.

Field rating

The average Umpiry Rating of the sixteen strongest players in the main draw, given as the share of this season’s fields it beats. The top sixteen, because in a large draw the weak end would outweigh the head; the overall rating, so that fields on different surfaces compare on one scale.

What counts

Singles main draws of one edition with at least sixteen rated players. Team events, exhibitions, juniors, doubles and cancelled matches are left out, and a rating is published once the season has twenty such fields.

How to read it

The line under the rating compares the draw with fields of the same level (“Stronger than 38% of WTA 250 fields this season”); with fewer than ten of them it gives the rank instead (“2nd strongest of 4”). Top 50 and Top 100 count entrants inside those places of the latest official rankings.

Title run

Once the tournament is over, the card shows how hard the champion’s path was: the opponents beaten, their average rating and the toughest win. Byes and walkovers are not counted as wins.

What it rewards

How often the two played, how close the head-to-head stayed (an even record counts for more than a lopsided one), and how much was at stake — finals and Grand Slam meetings add the most.

No current rating

Retired greats are ranked on the same footing as today’s players rather than fading from the list.

Head-to-head counting

Every completed meeting between the two players, Challenger and ITF included, walkovers excluded.

What it does not mean

It is a relative ranking and descriptive context only.

Tennis Records

The modern-era record book: the biggest achievements in singles since 2000.

Coverage

Titles, match wins, win streaks, season and surface records, and finals since 2000. ATP and WTA are counted separately and never merged into one list.

Eligible matches

Completed main-draw singles matches at tour level (Grand Slams, 1000, 500 and 250 events, the season finals) and the Olympics. Qualifying and team events do not count.

Retirements and walkovers

A retirement is a normal result. A walkover counts toward titles, finals and final win %, but not toward match wins, match win % or streaks, and it does not break a streak.

Seasons and streaks

A season follows the date of each event, so a tournament that starts in the last days of December belongs to the next season. Season records use completed seasons only. A win streak counts once it has ended, or after a year without a match (on that surface, for surface streaks).

Minimum sample

200 matches for career win %, 40 matches for season win % and 20 finals for final win %.

Ties

Players with the same value share a rank and the next rank is skipped (1, 1, 3); every tied holder is listed.

Updated

Every day. Families that can be checked against a finished career, such as Grand Slam titles, are published only when they match it.

Careers before 2000

Only their matches from 2000 onward count, so some totals are lower than official career totals. Missing history is not zero.

Data coverage

Coverage vs freshness

Data freshness is different from the player Freshness metric.

Data sources

Data is aggregated and cross-checked from multiple sources, then normalised so that players, tournaments and results line up across the platform.

Official ranking data

ATP and WTA rankings.

Match data

Results, scores, schedules and draws.

Point and game statistics

Used for Serve, Return, Clutch and Match Momentum.

Surface and tournament metadata

Surface, category, dates and draw.

When data is unavailable

This is a global Umpiry rule: missing data is never shown as 0.

Model limitations

Uncertainty

Tennis outcomes remain uncertain.

Context

Injuries, conditions and other context may not be fully captured.

Small samples

Small samples are less reliable.

History

Ratings depend on the historical data available.

Certainty

Model outputs should not be read as certainty.

What Umpiry does not infer

Umpiry copy stays factual. The data is never used to claim that a player is injured, tired, mentally weak or “choking”. Such facts appear only when confirmed by an external source.

Responsible play

Umpiry provides data and analysis for informational purposes only. It is an 18+ platform that supports responsible play and does not offer, accept or promote betting of any kind. If you choose to gamble, do so responsibly: set limits, never wager more than you can afford to lose, and seek support if betting stops being fun.