what happens if I lose to a lower-rated player?

Your rating drops more than it would for losing to someone at your level, because Glicko-2 expected you to win. The exact delta depends on the rating gap and your current deviation — new players with high uncertainty move further on a single result than established players do.

why are there separate leaderboards per language and board?

An English fixed-board game and a Dutch random-board game are genuinely different problems: different dictionaries, different premium layouts, different opening theory. Mixing the ladders would punish specialists and reward dabblers. Ratings should stay separated by language and board layout.

when does my rating actually update?

After covered rated online games, a few seconds after the analysis screen appears. The `updateStats` Cloud Function runs the Glicko-2 math server-side, writes the new rating and deviation, and pushes the change back to your profile. Local games and AI games never move your rating.

do AI games affect my rating?

No. AI opponents exist at several difficulties and are always labelled, but games against them are unrated. Rating changes come only from rated online games against other human players through the matchmaking queue or a friend invite with rating enabled.

what is the provisional rating?

For your first 10 rated games the rating is shown as Provisional — the Glicko-2 deviation is still wide, so each result swings your number more than it will later. After 10 games the rating is treated as established and future updates are smaller and more stable.

how hard is it to reach Laureate?

Laureate starts at 2300, which is 800 points above the 1500 starting line. Most players who reach it do so after a long run of rated games on a single language × board combination — specialising matters, because each ladder is its own climb.

From Novice to Laureate: how the rating works

A competitive player’s guide to the ladder, the matchmaking window, and what actually moves your number.

Last updated April 17, 2026 · By Kurt Bijl

In short: WordSalvo uses Glicko-2 — the same rating family chess federations use — for rated online games. Everyone starts at 1500 and lands in one of ten named tiers from Novice up to Laureate. Matchmaking targets opponents inside a ±200 rating-point window, and leaderboards are kept separate per language and per board layout.

A stylised rating graph climbing from the Novice tier at the bottom to Laureate at the top, with ten tier bands stacked along the vertical axis.

Why Glicko-2, not Elo

Rated online games in WordSalvo settle through a Glicko-2 update run inside the `updateStats` Cloud Function. Glicko-2 is the successor to Elo designed by Mark Glickman; unlike plain Elo it tracks a second number alongside your rating — a rating deviation that measures how certain the system is about where you actually sit. A short deep-dive by its author lives at glicko.net.

The practical effect: a new player with a wide deviation moves faster than a veteran, and an upset against a much higher-rated opponent shifts your number more than a win against someone just below you. Your rating is labelled Provisional for the first 10 rated games while the deviation is still wide, then becomes established.

The ten named tiers

Your current rating drops you into one of ten named tiers. Everyone starts at 1500, which places you in Adept after the first provisional games. The tiers are intentionally wide enough that you are not bouncing between names every other match, and narrow enough that a real run of good play moves you up.

The top band, Laureate, begins at 2300 and has no ceiling. The entry band, Novice, runs from 0 up to 1149. The table below is the full ladder with cut-offs.

WordSalvo rating ladder

Tier	Rating range	Where new players land
Novice	0 – 1149	Early ladder or a rough calibration
Apprentice	1150 – 1299	Building stable results
Scholar	1300 – 1449	Below the 1500 starting line
Adept	1450 – 1599	The 1500 starting line sits here
Virtuoso	1600 – 1749	Winning more than expected
Savant	1750 – 1899	Strong rated play
Maestro	1900 – 2049	Consistent high-level results
Sage	2050 – 2149	Near the top of the ladder
Grandmaster	2150 – 2299	Elite rated play
Laureate	2300+	No ceiling — compete for the leaderboard

Ten stacked horizontal tier bands labelled conceptually from a thin bottom band up to a tall open-topped band at the top, each rendered in a slightly deeper shade of warm ink.

Matchmaking: the ±200 window

When you join the online queue, WordSalvo looks for the closest-rated opponent inside a ±200 rating-point window. That is the close-match target — tight enough that an Adept player will not be thrown against a Laureate on their first matched game, loose enough that the queue can still return a human at active hours.

If nobody shows up inside the window, the band widens gradually with wait time rather than snapping open. You can always leave the queue and start an AI game or a friend match instead — AI opponents are clearly labelled and do not affect your rating.

Leaderboards per language and per board

WordSalvo goes live with English and Dutch dictionaries and 2 board layouts (the fixed custom premium pattern and the random-board mode that re-rolls multipliers each game). Ratings and leaderboards are kept separate for every language × board combination — so your English fixed-board rating is independent of your Dutch random-board rating.

The reason is simple: a bingo-heavy English game is a different problem from a Dutch game, and a random board rewards different opening theory than the fixed one. Mixing the ladders would punish specialists. Pick a lane, or climb a few in parallel.

Tournaments, ratings, and rivalries

On top of the global rating there is optional tournament play. Your rating still helps set the competitive context, while tournaments give players a more focused reason to play a sharper stretch of games.

Head-to-head against a recurring opponent is tracked as a rivalry, separate from rating. The game-over screen surfaces a rivalry card with your lifetime record — a Laureate can still be 1-5 against a specific Scholar, and the site keeps score of that honestly.

How to climb

Finish your first 10 rated games quickly to burn off the Provisional tag — the system moves you fastest while your deviation is wide. After that, your rating updates by a few points per match based on the gap between actual and expected result, so chase the games where Glicko-2 thinks you should lose. Beating higher-rated players is where real ladder movement happens.

Use the post-game analysis to find the moves you left on the table — the Brilliancy Score is the share of optimal points you actually captured, and climbing it is a better long-term proxy for rating growth than chasing any single win. A rating graph on your profile shows the curve so you can see whether you are trending or plateaued.

A horizontal rating strip with dozens of small tokens distributed along it and a bright highlighted band showing a ±200 window around a central candidate.

Frequently asked questions

what happens if I lose to a lower-rated player?: Your rating drops more than it would for losing to someone at your level, because Glicko-2 expected you to win. The exact delta depends on the rating gap and your current deviation — new players with high uncertainty move further on a single result than established players do.
why are there separate leaderboards per language and board?: An English fixed-board game and a Dutch random-board game are genuinely different problems: different dictionaries, different premium layouts, different opening theory. Mixing the ladders would punish specialists and reward dabblers. Ratings should stay separated by language and board layout.
when does my rating actually update?: After covered rated online games, a few seconds after the analysis screen appears. The `updateStats` Cloud Function runs the Glicko-2 math server-side, writes the new rating and deviation, and pushes the change back to your profile. Local games and AI games never move your rating.
do AI games affect my rating?: No. AI opponents exist at several difficulties and are always labelled, but games against them are unrated. Rating changes come only from rated online games against other human players through the matchmaking queue or a friend invite with rating enabled.
what is the provisional rating?: For your first 10 rated games the rating is shown as Provisional — the Glicko-2 deviation is still wide, so each result swings your number more than it will later. After 10 games the rating is treated as established and future updates are smaller and more stable.
how hard is it to reach Laureate?: Laureate starts at 2300, which is 800 points above the 1500 starting line. Most players who reach it do so after a long run of rated games on a single language × board combination — specialising matters, because each ladder is its own climb.