Chess rating system

How rating works on ChessHere.

ChessHere uses Glicko-2 — a modern rating system that's more accurate than classic Elo. Every rated game updates your rating instantly, and the math is open and transparent.

1500
Starting rating for every new player.
5
Independent rating pools, one per time control.
Per-game
Your rating updates the moment a game finishes.
FIDE scale
Numbers feel familiar to anyone who knows Elo.
Why Glicko-2

A more accurate rating, faster.

Classic Elo treats every player the same: every win or loss moves your rating by a fixed amount, regardless of how long you've been playing or how confident the system is in your number.

Glicko-2 is smarter. It tracks how confident the system is in your rating, and adjusts faster when it isn't sure — and slower when it is. New players climb to their real strength quickly. Established players don't get yanked around by a single lucky streak.

It's the same family of system used by major correspondence chess federations and the closest thing to a "FIDE rating, but designed for the internet" that exists today.

The three numbers

Behind every rating, three values.

You only see the rating itself, but the system tracks three numbers per player, per time control. Each one tells the system something different.

1

Rating

The number you see. Higher means stronger. Starts at 1500 for everyone.

Same scale as FIDE / Elo.
2

Rating Deviation (RD)

How sure the system is about your rating. New players start at 350 (very uncertain); active players settle around 30–80 (very confident).

High RD → bigger rating swings per game.
3

Volatility

How likely your true strength is changing. Starts at 0.06. It drifts up when you have surprising results, and down when your results are consistent.

Captures sudden improvement or decline.
Five rating pools

A rating per time control.

Bullet skill and classical skill aren't the same thing. Your rating in each time control is tracked separately, with its own RD and volatility, so a brilliant blitz run doesn't change your daily-chess number and vice versa.

Bullet
Under 3 minutes per side.
Blitz
3 to 8 minutes per side.
Rapid
8 to 25 minutes per side.
Classical
25 minutes or more per side.
Correspondence
Days-per-move games — daily chess.
How a game changes your rating

Three things matter, every game.

A

Who you played

Beating a higher-rated opponent gives you more points than beating a lower-rated one. Losing to a much lower-rated opponent costs you more than losing to a much higher-rated one.

B

How sure the system is about you

A high RD (new player, or returning after a long break) means each game can move your rating dozens of points. A low RD (regular player) means it moves only a few points per game.

C

How sure the system is about your opponent

A win against an opponent the system is uncertain about (high RD) carries less weight than a win against an opponent it knows well. The system trusts more confident ratings more.

Before the game

On every challenge dialog and seek, ChessHere shows you exactly how your rating would change for a win, draw, or loss — calculated against your specific opponent. No surprises after the game.

Provisional ratings

A new player starts with RD = 350 — the system has no idea how strong they actually are. Their rating is shown with a ? to mark it as provisional.

The "?" disappears once the system has seen enough games to be confident. Until then, expect big swings — and try to find your real strength quickly. The faster you play, the faster the "?" goes away.

Inactivity drift

Skill changes when you don't play. After a long break, the system isn't as sure about your rating, so your RD goes back up — capped at the starting 350.

Your rating doesn't move during a break, only the confidence around it. Come back, play a few games, and the system will quickly re-establish your real number.

If you know FIDE / Elo

What's the same, what's different.

  FIDE / Elo Glicko-2 on ChessHere
Starting rating ~1400 unrated estimate, but no formal start. 1500 for everyone.
Confidence Not tracked. Same K-factor for everyone. Tracked per player as RD — bigger swings when uncertain, smaller when confident.
Update timing Periodic (often monthly). After every rated game.
Per time control Yes (FIDE separates standard / rapid / blitz). Yes — bullet, blitz, rapid, classical, correspondence.
Rating scale ~1000 (beginner) to 2800+ (world top). Same scale — numbers feel familiar.
Inactivity Some federations apply rating decay. RD increases (uncertainty grows). Rating itself doesn't decay.
Why we chose Glicko-2

Fairer for new players, kinder to regulars.

  • New players reach their true rating fast.
    High RD at the start means each game moves the rating a lot. Fewer games to find your level.
  • Regular players don't get whiplash.
    Low RD means a single bad game won't tank a hard-earned rating. Consistency is rewarded.
  • Sudden improvement is recognised.
    Volatility responds to surprising results, so a player who's improving fast climbs faster than they would under classic Elo.
  • The math is open.
    Glicko-2 was designed by Professor Mark Glickman and is fully published. Anyone can verify exactly how a rating was calculated.

Ready to play a rated game?

Sign up free and start building your rating. Every rated game counts — in every time control.

Frequently asked questions

  • How does rating work?

    ChessHere uses Glicko-2, a rating system that tracks both your skill (the rating number) and the system's confidence in that estimate (rating deviation). After each rated game, both numbers update based on the result, your opponent's rating, and the time since your last game. Each time control has its own pool: bullet, blitz, rapid, classical, correspondence, and puzzles all rate independently.

  • What is a provisional rating?

    Your rating is provisional during your first games in a pool, while the system has high uncertainty about your skill. Provisional ratings move in larger steps as the system zeroes in. Once you've played enough games — usually fifteen to twenty — your rating deviation drops below the threshold and the rating is no longer marked provisional. Provisional and established ratings count equally for matchmaking.

  • What is rating deviation?

    Rating deviation (RD) is the system's measure of confidence in your current rating. A new player has high RD — the system isn't sure where they belong. After lots of recent games, RD drops, meaning your rating is reliably calibrated. RD also rises when you don't play, reflecting that your skill may have changed. The Glicko-2 paper at glicko.net describes this in detail.

  • Does my rating decay if I stop playing?

    Your rating number doesn't drop, but your rating deviation grows over time when you don't play. Higher deviation means the system is less confident in your rating, so when you return, your first few games will move your rating in larger steps until the deviation drops back. There is no automatic rating-decrease for inactivity on ChessHere.

  • Why does ChessHere use Glicko-2 instead of FIDE Elo?

    FIDE Elo is designed for tournament chess with a small fixed pool of titled players. Glicko-2 is better suited to the millions of online games per day, where players have wildly varying activity levels and skill changes can be fast. Glicko-2 tracks rating deviation, which lets the system adapt to new and returning players cleanly. The two systems use different numbers — your ChessHere rating isn't directly comparable to a FIDE rating.