Padel Ratings: The Case Against One Global Number

Quick Answer: The problem with one global rating is not accuracy, it is scope. A single number has to describe you to strangers, gate your matches and settle your crew's arguments at once — so it hides its uncertainty, prices your context away, and starts changing how you play. Scoped numbers with visible workings do the same job without the harm.

A rating is not a fact about you. It is a summary written for somebody who has never seen you play.

That is not an accusation. It is a job description, and it explains almost everything players find maddening about their level. This page argues that the single global number is the wrong shape for amateur padel — and it argues it the honest way, which means starting with everything the global number gets right.

First, the case for the number

Before it is criticized it should be credited, because it solved something hard.

Open matches only work because a shared currency exists. Four strangers can be put on a court together at 19:00 on a Tuesday and have a real match, and that is a genuinely remarkable piece of infrastructure. Before platform levels, the alternative was a coach who knew everybody, or a WhatsApp group and hope. Every scale that gets criticized — 0 to 7, 1.0 to 7.0, the club's A to D — exists to answer one unavoidable question: will this be a good match?

The number also does honest work for competition. Draws have to be seeded. Grades have to have entry criteria. A national body cannot run a tour on vibes.

So the argument here is not that ratings are bad. It is that one number cannot hold everything we have asked it to hold.

One number, three jobs

Look at what your level is simultaneously required to be.

Those requirements pull against each other. A passport wants to sit still; a verdict wants to move the week you improve. A gate has to be defended against manipulation, which means the system cannot be fully transparent about how it works — and an unexplainable number is exactly what nobody wants as a verdict on themselves.

Every platform lands somewhere in that compromise, and every player experiences the compromise as unfairness aimed personally at them. It is not personal. It is arithmetic under conflicting instructions.

What compression costs

A single number is a compression of a season into one decimal, and compression is lossy in specific, knowable ways.

It loses who you played. A win over the strongest pair at your club and a win over four beginners are the same row in your record until a system prices them. Good systems do price them, which is why a 70% win rate can be worth less than a climb — the argument in win rate versus rating.

It loses by how much. Platforms have to draw a line on what counts as a result at all, and the lines are visible in their own documentation. Playtomic's page on how its level system works sets out its validity rule: a set needs one side on 6 or 7, and if nobody reached 7 there must be a two-game gap — so a 6-4 counts and a 6-5 does not. A 6-5 is one of the best matches padel produces. As information it is discarded, for a defensible reason: a rule that admits every scoreline admits every made-up scoreline too.

It loses who you played with. In 2v2 your result is priced against your pair, not against your afternoon. Playtomic's own explainer describes the algorithm averaging the level of each team and comparing the expected outcome with the actual one. That is the right way to do it and it still means a night carrying a much weaker partner is charged to you at full price — the effect worked through in level gaps in doubles.

It loses how sure it is. This is the deepest one. Your level is displayed as a number; the system's confidence in that number usually is not. Playtomic computes a reliability score that decides how far each result moves you, and states plainly that a player cannot see it changed, adjusted by them or by customer support. So two players show each other a 3.5 that means completely different things: one is an estimate with ninety matches behind it, the other is a guess from a questionnaire.

Nobody is being deceived. Something true is simply being left out, and it is the part that would explain everything.

Then the number starts policing the play

Here is the cost that is not technical.

Attach consequences to a measure and people will manage the measure. This is not a padel problem or a character problem; it is what measures do. In padel it shows up as recognizable behavior:

• Not entering a match you might lose, because the number is close to a round figure.
• Choosing the partner who protects the average over the partner who makes the night fun.
• Under-stating your level so the matchmaking is comfortable — the subject of sandbagging in padel.
• Treating a permanent decision as an escape hatch: lowering your level manually is a one-way door on Playtomic, and people still reach for it.

The strongest evidence that this is real comes from the platforms themselves. They ship a casual match option: a way to play padel that does not count. Playtomic's documentation is explicit that only competitive matches affect your level. That toggle is an admission built into the product — some padel should not be measured, and the people who built the measure know it.

Think about what the toggle implies. If a number needs an off switch to keep the sport enjoyable, the number's scope was too wide.

The alternative is not fewer numbers

This is where the argument would usually end with a shrug about playing for fun. That is the wrong conclusion, and it insults everyone who genuinely likes counting things. Padel is competitive. Keeping score is the point.

The fix is not silence. It is scope — plus two properties that any number can have and most do not.

1. Say who the number is about. A rating scoped to the four people who play every Tuesday has one job instead of three. It does not have to describe you to strangers, so it never has to be defended, and there is nothing to gain by managing it. Nobody sandbags a number that only buys bragging rights among witnesses. That is the whole argument in should your friend group track its own rating, and the reason it works is boring: consequences removed, pathology removed.

2. Show the work, every time. Any move should survive one sentence spoken out loud to the people who were on court: "you were expected to take 40% of the games and you took 55%, so you gained." A number that cannot be explained in a sentence is not a verdict, it is an announcement. Run one of your own results through the padel Elo calculator, then the same result with a different scoreline, and you can watch the explanation change — including the part that offends everyone at first, which is that beating a pair you were 85% likely to beat is worth almost nothing.

3. Show the doubt. Uncertainty is not an embarrassment to be hidden; it is half the information. This is settled engineering, not an opinion: Microsoft's TrueSkill ranking system keeps an explicit uncertainty alongside every skill estimate, because "maintaining an uncertainty allows the system to make big changes to the skill estimates early on but small changes after a series of consistent games has been played". A number that says provisional for its first ten to fifteen sets earns trust that a confident-looking guess spends. And a rating should grow less certain when you go on holiday, never lower — punishing absence is how you lose a player.

Scope, shown work, visible doubt. None of it requires abandoning ratings. All of it requires giving up the fantasy that one number can serve every purpose at once.

What that looks like on a Tuesday

Four of you, one court, three matches, everyone partners everyone.

The board that comes out of that night can hold several true things at the same time: who is playing best, who has improved most in a month, which pair is quietly unbeatable, and who is 7-3 against whom — a head-to-head record being the one number anyone quotes voluntarily. Publish several boards and four different people can be winning something. Publish one ranking and its bottom half stops coming.

That is what we build Rivals for: a margin-aware rating that lives inside your crew and nowhere else, honest about how sure it is, and able to tell you in one sentence why it moved — alongside head-to-head records, belts, a records book and a year in review. Free is one crew of up to four players, with nothing inside it metered and no ads. Keep your platform level as well. It is a passport, and passports are useful.

But keep it in its place. Your global level is a stranger's estimate of you, built for a job you barely care about. The number that matters is the one earned in front of three people who will remember it.

Under the lights there is nowhere to hide — and among friends, that is exactly the point.

For the full map of levels, scales and what each is worth, start with the padel levels and ratings guide.

FAQ

Are padel rating systems flawed?

Less flawed than over-loaded. A single global level is asked to be a passport for strangers, a gate for competitive entry and a verdict on your own game at the same time, and those three jobs need different things. Most complaints about accuracy are really complaints about one number carrying three purposes.

Why does my padel level feel unfair?

Usually because it hides the context that would justify it: which opponents you faced, by what margin, with which partner, and how confident the system actually is. Platforms price all of that internally and display only the result, so a defensible move arrives looking arbitrary.

Should I stop caring about my padel level?

No — use it for what it is good at, which is getting matched with strangers at roughly the right standard. What it is bad at is telling you whether you are improving or who is best among your friends. Keep a second, scoped number for that.

Is a private rating better than a global one?

For ranking a fixed group of friends, yes, on three counts: the data is denser, nobody has an incentive to manipulate it, and every move can be explained to the people who watched the match. It is worthless outside the group, which is precisely why it stays honest.

What makes a rating trustworthy?

Three things. A stated scope, so you know who it is about. An explanation for every change, in one sentence. And a visible level of uncertainty — provisional while the system is still learning you, and growing less certain rather than lower when you stop playing.