How d1R works

d1R

Skill Rating

Cayman Protocol · Glicko-2 · No time decay

What is d1R?

d1R is your skill rating - a calibrated estimate of how good you are, built on Glicko-2 math. It answers: "can you win?"

Every confirmed match updates your d1R based on the gap between what the system expected and what actually happened. Upsets move it more than predictable wins. Losses to stronger players barely move it; losses to weaker players move it more.

Key point: d1R is not your ranking points. Those are tracked as d1P, which drives your ladder position. d1R measures ability; d1P measures recent activity + achievement.

d1R vs d1P - two different measures

d1R (Skill)

Ability-based
Calibrated skill estimate
No time decay
Changes slowly
Answers: "how good?"
Like chess Elo, tennis UTR

d1P (Ranking points)

Activity-based
Rewards playing & winning
Decays over 365 days
Changes constantly
Drives your ladder position
Like ATP points

Neither replaces the other. An elite active player has both high d1R and high d1P. High d1R + low d1P = skilled but on a break.

The Cayman Protocol - three rules

d1R is computed by Cayman Protocol v3.2 (algorithm slug cayman_v1),a Glicko-2 variant with three rules layered on top. Glicko-2 math was invented by Mark Glickman (MIT) - the same foundation used by the US Chess Federation.

1 · Winners never drop

Win a match, your d1R cannot go down. Even against much weaker opponents. The worst outcome for winning is that your rating holds steady.

2 · Losers can still gain

Lose a close match against someone much stronger and you may gain d1R. You outperformed expectations - the system rewards that, not the final scoreline.

3 · Partner disparity dampening

In doubles, if one player is far stronger than the other, the rating delta is dampened. The strong partner can't pump rating by carrying a weak one; the weak partner can't hide behind the strong one.

What moves your d1R

1 · Expected vs actual score

Glicko-2 predicts the score before the match based on both players' ratings. Your d1R moves by the difference between expected and actual - beating someone you should beat barely moves it; upsetting a stronger player moves it a lot.

2 · Rating deviation (RD)

RD is how sure the system is about your rating. A new player has high RD (wide confidence band) so each match moves them more. After many matches, RD shrinks and moves get smaller. Reliability grows with matches.

3 · Trust multiplier

Match-type weight x confirmation status x venue verification x integrity flags. Fully confirmed matches at verified venues count the most; unconfirmed or flagged matches count less. Stored on rating_history.trust_mult.

4 · Format subscores

Your primary d1R is overall (doubles + singles combined). Separate subscores for doubles and singles are maintained in parallel. Leaderboards and profiles default to overall.

Rating tiers

Default tier thresholds. Clubs can override via tier_configs.

Beginner0.0 – 2.5

Developing2.5 – 3.0

Intermediate3.0 – 4.0

Advanced4.0 – 5.0

Elite5.0+

Why doesn't d1R decay?

Skill doesn't evaporate. Stop playing for a year and you're still roughly the same player - your RD grows instead (the system is less sure about the number), so your first few matches back will move your d1R more until it re-calibrates. Ranking points (d1P) penalize inactivity, not skill.

The trust multiplier

Raw Glicko-2 deltas are multiplied by a trust factor that blends: match-type weight (open play vs league vs tournament), confirmation status (fully confirmed vs one-sided), venue verification (staff-verified adds a boost), and integrity flags. Every breakdown is stored on rating_history.trust_breakdown - fully auditable per match.

How your d1R changes day-to-day

New confirmed match: d1R updates as soon as the match is fully confirmed. Unconfirmed matches carry a reduced trust multiplier - some don't count at all until an independent signal confirms them.

No nightly drift: Unlike d1P, there's no decay job. If no matches happen, your d1R stays put - only the RD around it slowly grows.

Retroactive recomputation: If a match is edited, unconfirmed later, or flagged for integrity, the rating chain is recomputed from that match forward. Your current d1R always reflects the corrected chain.

Where you see d1R

Your dashboard tile: current d1R and tier. Tap to open this explainer.

Leaderboards (all levels): every leaderboard has two tabs - Ranking · d1P (default) and Skill · d1R.

Your profile + stats: d1R appears alongside tier, reliability, rating history, and confidence band (low / high).

Match entry preview: before you confirm a score, preview_match_ratings shows exactly what your d1R will move to.

The big picture

d1R measures ability. d1P rewards activity. Together they tell the full story. A 4.8 d1R player who logs 4 matches a year has genuine skill but no ranking. A 3.2 d1R grinder who plays 150 matches a year climbs the d1P leaderboards without needing elite skill. Both numbers matter - neither replaces the other.

Questions?

Why doesn't d1R ever go down when I win?

Rule #1: winners never drop. Even if you beat someone much weaker (where the expected outcome was already a win), your d1R holds steady. This prevents punishment for showing up against a mixed field.

Why is d1R different from DUPR / UTR / Elo?

DUPR is also Glicko-based but uses different trust weighting and no separate ranking axis. UTR is proprietary. Elo is a pairwise system without RD. Cayman Protocol layers dink.one's three rules on top of Glicko-2 math, plus trust-weighted deltas specific to our confirmation policy.

Why does d1R not decay over time?

Skill doesn't evaporate - if anything, trained ability persists. What's missing from d1R after an inactive period is certainty, not skill: your RD grows while you aren't playing, so your next few matches will move your d1R more. Ranking (d1P) is where inactivity penalizes you.

How many matches until my d1R is reliable?

Depends on RD. A new player hits the "reliable" threshold around 10-20 confirmed matches against a range of opponents. Playing only against the same partner slows this down (RD shrinks slower).

Can rating be retroactively recomputed?

Yes. If match data is edited (score corrected, match unconfirmed later, integrity flag raised), the full rating chain is recalculated from that point forward. Opponent strength for d1P is snapshot-frozen so d1P history is stable, but d1R always reflects the current corrected chain.