Stats & Rankings

Understand how your performance is tracked and what your statistics mean.

Your Rating (Glicko-2)

Tabletop League uses the Glicko-2 rating system to measure player skill. Glicko-2 is a modern successor to Elo: it tracks not just how good you are, but how confident the system is in that estimate — so a newcomer and a veteran on the same number are treated very differently.

How It Works

Every player starts at 1500.
Ratings are per game system — your Unmatched rating is separate from your other games.
Your rating updates after each rated match. Beating a stronger opponent moves you up more than beating a weaker one; the same logic applies to losses.
Only matches in competitions Tabletop League has marked Rated affect this rating. You'll see a "Rated" badge on those competitions.
Forfeits and byes never move your rating. Only matches you actually play count — a walkover win gains you nothing, and a forfeit loss from a scheduling conflict or vacation costs you nothing. (Competition standings are a separate system: forfeits still cost standings points there.)

Rating Deviation (the "± number")

Your rating is shown as 1650 ± 85. The second number is your Rating Deviation (RD) — how uncertain the system is about your true skill:

High RD (new or returning players) → the rating can swing a lot per game.
Low RD (players with a steady recent record) → the rating is settled and moves in smaller steps.
RD shrinks as you play rated matches, but never below 60: the system always keeps a little room to move.
RD grows while you're away from rated play, one step for each full 30 days, and your rating number stays exactly where it was. See Time away from rated play.

Provisional ratings

While the system is still learning your skill, your rating is marked provisional. A rating stays provisional until you've played at least 10 rated games and your RD has dropped below 200. Because RD grows during time away, a long break can make a settled rating provisional again. Provisional ratings are shown with an outline badge and (provisional) label so others know not to read too much into them yet.

Volatility

Glicko-2 also tracks volatility — how erratic your recent results have been. A string of unexpected upsets temporarily makes your rating move faster; a run of consistent results settles it down. This happens automatically; there's nothing you need to do.

How your rating is calculated

The settings behind your rating, and where they differ from textbook Glicko-2:

Starting values. Everyone starts at a rating of 1500, an RD of 350 and a volatility of 0.06. The system constant (τ), which limits how fast volatility can change, is 0.5.
Limits. RD always stays between 60 and 350, and a rating never drops below 100.
One update per match, straight away. Your rating changes as soon as a rated match's result is in. Textbook Glicko-2 collects results into rating periods (a week, say) and updates everyone at the end of each one; here every rated match is its own rating period.
The match result counts, not the games. A 2-0 and a 2-1 win move your rating by the same amount, and a best-of-three or best-of-five counts as a single result. A drawn match counts as half a win for each player. Why: in games with a draft, a player may deliberately lose a game they've already won to steer the next draft, so one game on its own is a poor read of skill.
What isn't rated. Besides forfeits and byes, multiplayer pods (such as Blitz) and matches nobody won in a game that can't end in a draw don't count.
Time away. For each full 30 days without a rated match, your RD grows by one fixed step (39 on the ± scale), up to the 350 limit. Your rating number never changes while you're away. This started on 1 November 2026: time away before that date doesn't count. Details below.
Leaving the ranked board. You drop off the ranked leaderboard after 12 months without a rated match. Your rating is kept.
Corrections. If a result is reopened or changed, its rating change is undone, and the corrected result is rated when it's reported again.

Time away from rated play

Will I lose points for not playing? No. Your rating number never drops because you took a break. What changes is your ± (RD): for each full 30 days without a rated match, it grows by one step. Under 30 days changes nothing.

Why. Time away isn't evidence that you've got worse. It only means the system knows less about where you are now. A bigger ± tells the system to trust your first results back a bit more, so you get to your real level quickly, in either direction. This is how Glicko-2 is designed to work (see Glickman's paper under Further reading).

The schedule. Your ± after time away is worked out from the ± you had after your last rated match:

new ± = √( old ±² + months × 39² ), never above 350

Here "months" means full 30-day periods. The steps add up as squares, not straight sums: a settled ± of 60 becomes about 72 after one month (√(60² + 39²)), not 99. The longer you're away, the smaller each extra month's change, until it reaches 200 at about two years.

An example. A settled player at 1700 (± 60) comes back and beats another settled 1700:

Time away	± going in	First win back	Opponent's loss
Under 30 days	60	+10	−10
1 month	72	+14	−10
3 months	90	+22	−10
6 months	113	+33	−10
12 months	148	+53	−9
24 months	200 (provisional)	+86	−9

Away 6 months, the first win back is worth +33 instead of +10. A first loss back costs the same amount the other way.

Fairness for your opponent. Beating a returning player moves your rating slightly less (−10 becomes −9 in the table above, and the same for a win), because the system is less sure how strong they are right now.

Provisional again. If your ± reaches 200, your rating is marked provisional again and you leave the ranked leaderboard until a few rated matches settle it. Your rating is kept. With these settings that takes about two years away.

How long it takes to settle. One rated match takes a returning player straight back out of provisional (± 200 becomes about 174). After that the ± keeps shrinking with each match: roughly 157, 144, 134 and 126 over the next four, and about 100 by the tenth.

How we differ from textbook Glicko-2. Textbook Glicko-2 grows RD by your volatility each rating period, which for a monthly period barely registers (a whole year away would move a settled ± from 60 to about 70). We use a fixed monthly step instead, the way the original Glicko system does, sized so that about two years away makes a settled rating provisional again. See Glickman's Glicko page.

When it started. Time away counts from 1 November 2026. Breaks before that date don't count, and ratings earned before it were not recalculated.

Organizer ratings. The same rule applies to an organizer's Glicko-2 ratings (their own RD limit applies). Organizer boards that use Elo are not affected.

You'll see this wherever it shows: a ± that has grown since your last rated match opens a short explanation when you tap it, and the match history says when a big move came from a return after time away.

Questions about time away

Why did my ± go up when I didn't play? Each full 30 days without a rated match adds one step to it. Your rating itself didn't change. See the schedule above.
Did I lose rating points for taking a break? No. Time away only changes your ±, never your rating number.
Why did I get so many points for one win? If you were away for a while, your ± had grown, so your first results back count for more. When the effect is noticeable, the match history says so next to that match. It works the same way for a loss.
Why did my rating move less than usual against someone? Your opponent was coming back from time away, so the system was less sure how strong they were and gave that result a little less weight. When the effect is noticeable, your match history says so next to that match.
Why did I disappear from the leaderboard? Either your ± passed 200 (about two years away), which makes your rating provisional again, or you haven't played a rated match in 12 months. Either way your rating is kept, and rated matches bring you back.
How do I get my ± back down? Play rated matches. One match takes a ± of 200 to about 174, and about ten bring it near 100.
Does a break from before November 2026 count? No. Time away counts from 1 November 2026 only.
Does this apply to organizer leaderboards? Yes for organizers that use Glicko-2 ratings, with their own ± limit. Elo boards aren't affected.

Why not plain Elo?

Plain Elo treats a 5-game player and a 500-game player identically once they share a number, and never expresses doubt. Glicko-2's RD and volatility mean new players climb (or fall) to their real level quickly without distorting established ratings — a better fit for tabletop, where players come and go between events.

Leaderboards

View rankings on the Leaderboard page:

Global - All players across the platform
Game System - Rankings within a specific game
Regional - Players in your area
Organization - Members of a specific group

Organizer ratings

Some organizers run their own rating leaderboard built only from the matches played at their events — a "house ranking" for their scene, separate from your platform-wide rating above. Every competition that organizer runs in that game counts, unless they've left it out, so the matches you've played at their events count even if the competition has no Rated badge. Members-only competitions and friendlies don't count. If you've played at an organizer that has ratings enabled, your standing there shows up in the Organizer ratings section of your profile, and on that organizer's Ratings page. Organizers can read how to set this up in Player Ratings.

Statistics Dashboard

Your profile includes detailed statistics:

Match History

Total games played
Win/loss record
Win rate percentage
Recent form (last 10 games)

Fighter/Character Stats

Most played characters
Win rates by character
Character usage trends

Character Statistics & Tier Lists

Beyond your personal stats, each game's Stats page (e.g. /games/unmatched/stats/characters) shows community-wide character performance with three views:

List view

Each character card shows total plays and raw win rate across all tracked competitions.

Matchup matrix

The 2-D matrix shows row-character win rate against each column-character. Cells with fewer than 10 recorded games show an italic estimate that combines the Bradley-Terry prediction with the cell's own record (see How matchup estimates work). Cells with 10 or more games show the observed win rate. Tap any cell to see how it was worked out and its per-map breakdown.

Friendly games in character stats

Friendly games, shared or private, count toward character stats right away. Players are never named from a private friendly.

Tier list

The Tier list tab ranks characters by Bradley-Terry strength into S / A / B / C / D / F bands (top 5% / 15% / 30% / 30% / 15% / bottom 5%). Characters with fewer than 10 games are hidden until they have enough data to rank. Unofficial versions of characters, such as alternate-art editions, are left out of the tier list, as they are from the character grid.

What Bradley-Terry means

Raw win rate falls apart on sparse data: a 3-2 record reads as "60%" with the same weight as a 60-40 record across 100 games. Bradley-Terry fits a single strength rating per character across the full matchup graph, so a character's rating reflects not just who they beat but how strong those opponents are. It also fills in matchups that never happened by propagating evidence through common opponents.

You'll see Bradley-Terry-derived "Expected vs avg" win rates on each character's detail page, and Counter / Favored / Even / Unfavored / Countered labels in the matchup table. For matchups with fewer than 10 recorded games, those labels come from the estimate below, not from the rating alone.

How matchup estimates work

A 1-5 record is too few games to trust on its own. A rating on its own ignores what actually happened in this matchup. So for any matchup with fewer than 10 recorded games, the win rate you see is an estimate that uses both:

the Bradley-Terry prediction from the two characters' overall ratings, counted as if it were 5 games, and
the matchup's actual record.

The formula is (prediction × 5 + wins) ÷ (5 + games).

Worked example. Character A is predicted to beat Character B 70% of the time from their overall ratings, but A is 1-5 against B. The estimate is (0.70 × 5 + 1) ÷ (5 + 6) = 41%, so the matchup reads "Unfavored". With no games yet you see the model's 70%; once the matchup reaches 10 games you see the record on its own.

Estimates are shown in italic with a ~. Tap any italic value (or matrix cell) to see its working.

What Bradley-Terry can't see

One number per character. The model has no idea that one character struggles against another specific character. That's why matchups use the estimate above, not the rating alone.
Characters, not players. A character played mostly by strong players rates higher than it would in average hands.
Concentrated games count fully. If one pairing is played very often, for example a group working through a single box set, that record moves both characters' overall ratings.
Draws count as half a win for each side.

The tier list ranks overall strength. For a specific matchup, trust the record once it reaches 10 games.

Weighted Win Rate (WWR)

Weighted Win Rate (WWR) is a per-competition character stat that credits each win by the quality of the character it beat — closing raw win rate's blind spot of counting a win over the field's strongest character exactly like a win over its weakest. Each win earns between 0.5 and 1.0 credit, scaled by the beaten character's match win rate in that competition (that rate excludes the two characters' head-to-head games, and thin records are smoothed toward the 0.5 average); a loss earns 0. A character's WWR is the average credit it earns per game.

Because credit maxes out below a full point, WWR is not comparable to raw win rate: a perfect record against merely average opposition lands near 0.75, not 100%. Use it to compare characters within a single competition, and rely on Bradley-Terry when you want to compare strength across different competitions.

Map Statistics

Performance by map
Win rates on each stage
Map preferences

Head-to-Head Records

Your record against specific opponents
Common matchups
Historical results

Tracking Progress

Use the stats to identify:

Strengths - Characters and maps where you excel
Weaknesses - Areas that need improvement
Trends - Whether you're improving over time
Matchups - Opponents you struggle against

Privacy

You control who sees your detailed statistics in your profile settings.