Radical transparency
How your rating works
No black box. This page is the whole algorithm, in plain English — and every rating change you ever receive comes with its own one-line explanation.
Your number and the ±
Your rating is a number like 1540, shown alongside the familiar scales (≈ NTRP 3.5 / UTR ~3.7) so it's easy to orient. It always comes with a ± — our honest uncertainty. A new player might be "1540 ± 150" (we're still learning your game); a regular is "1540 ± 70". In validation, the true level sits inside the ± range 97% of the time.
Where you start
You're seeded from the best evidence available, and we're honest about how much we trust each kind:
| What you give us | How much we trust it |
|---|---|
| Connected UTR account | Strong — you start nearly settled |
| UTR screenshot (we verify it) | Strong, scaled by UTR's own "reliability %" |
| Self-reported NTRP/UTR | Moderate — self-ratings run about a half-step optimistic on average, and we adjust for that |
| Our placement questionnaire | Loose — your first matches will move you quickly |
A wrong placement is not a life sentence. In our validation, a player seeded a full NTRP level wrong is typically back in the right band within about 10 matches — and can watch it happen.
How a match moves your number
Every result implies a level. Win 12 of 19 games against a 1520-rated opponent and you played like roughly a 1610 that day. Your rating moves partway toward what the result implies:
- A big step while we're still learning your level — new and returning players move fast.
- A small step once you're settled — one bad day can't wreck an established rating.
- A smaller step when your opponent's rating is unproven — beating a brand-new player teaches us less.
Game score matters, not just who won. Winning 6-2 6-3 as the favorite is expected; winning 7-6 7-6 means you underperformed the expectation — the favorite can lose points in a narrow win. That's what makes sandbagging pointless: your games give you away every match.
The fine print, also in plain English
- No single result is fully trusted. A score that implies an absurd gap (injury, retirement, someone's worst day of the year) is discounted automatically.
- Repeat opponents count less. Your 2nd match against the same person in a month counts half; the 3rd a quarter; beyond that it's just for fun. Trading wins with a friend can't farm points.
- Casual (free-tier) matches count at half weight — they keep your rating warm, but the ladder is where it's earned.
- Walkovers and no-shows never touch ratings. They're handled as conduct, not results.
- Your rating never decays and never resets. Away for six months? The number stays; only the ± widens, so your comeback matches move you faster. There are no seasons here.
Disputes
Think your placement is wrong? Every rating event is stored with its full inputs, so support can show you exactly why your number is what it is — and if something's genuinely off (a mis-entered score, a bad seed), it gets corrected with a visible, logged adjustment. No "trust us."