The ladders use Glicko-2, the system Mark Glickman designed as a successor to Elo. It is a better fit than win/loss or plain Elo for a club where people play a handful of matches a season.
Rating — the skill estimate. Everyone starts at 1500.
RD (rating deviation) — how sure we are. A newcomer starts at ±350; it shrinks as they play and creeps back up while they don't.
Volatility — how erratic their results are. Consistent players move in small steps; players who spring surprises move faster.
You carry a separate rating in each of the five divisions, because being strong at singles says surprisingly little about how you play mixed doubles.
Doubles is rated per player, but what you are measured against is the gap between the two teams. Each player is updated as though they had played one opponent chosen so that the gap they face equals the gap their team faced.
Say you're 1600 with a 1200 partner against two 1500s. Your team averages 1400 against their 1500, so you are scored as though you personally faced a 1700 player — win and you gain a lot. Your partner is scored against a virtual 1300, so you both share exactly the same team win expectation, and differ only in how far each of you moves. Carrying a weak partner past strong opponents pays properly; coasting behind a strong one does not.
Club formats vary, so the scoreline box takes whatever you actually played. Write side A's games first.
| Format | Type this |
|---|---|
| A single set — the usual challenge match | 6-4 |
| One set decided on a tie-break | 7-6 or 7-6(5) |
| Two sets and a deciding match tie-break | 6-4 4-6 10-8 |
| Full best of three | 6-4 3-6 6-2 |
| Eight-game pro set | 8-6 |
| Fast4 | 4-2 |
| Someone retired | 6-3 2-1 ret. |
| Walkover | w/o |
A deciding-set match tie-break counts as one game rather than ten, so a super-tie-break can't outweigh the two real sets before it. Every match counts the same regardless of format — a single set moves your rating as much as a three-setter. That's deliberate: one set is this club's normal format, and weighting short matches down would slow everyone's rating from settling, which matters most in a short season.
A score with the sets level and no decider
(6-4 3-6) is refused — that's an unfinished match, not a
format. Add the deciding set, or mark it ret.
Because who you beat is the whole story. Beating the #1 player moves you far more than beating the #20, and losing to a strong opponent barely costs you. A ladder ordered on win percentage rewards ducking tough opponents; this one rewards playing them.
A win is worth 80% of the credit outright; the remaining 20% is split by share of games won. So 6-0 6-0 scores 1.00, 6-4 6-4 scores 0.92 and 7-6 7-6 scores 0.91 — winning is what matters, but a thrashing is recorded as a thrashing. Retirements and walkovers ignore the margin entirely. A deciding-set match tie-break counts as one game, not ten.
Position is set by rating − 1×RD, not raw rating, so an unproven player can't top the ladder on one upset. Players are marked provisional until they have played 3 matches — or straight away, if their rating carried over from last season already well enough known (RD under 200).
Ratings carry into a new season with their uncertainty widened to at least 150, so where you finished seeds where you start without freezing it. Your first match in a new division starts from your singles rating held loosely (±300) rather than from scratch, so recruits settle in a few matches instead of a dozen. Every seed comes from a player's own results — nothing here encodes anyone's opinion of how good someone is.
Glicko-2 updates in periods rather than match by match, so results inside the same period are weighed against each other fairly rather than in the order they were entered. These ladders use 7-day periods. Ratings are always recalculated from the complete match history, so correcting an old result fixes everything downstream of it automatically.