Chessdigi.comFast chess. Learn from your own games.
Online Chess

Online Chess Ratings Explained: Elo, Glicko and Why Your Number Differs Between Sites

How chess ratings are calculated, what rating deviation and provisional ratings mean, why sites disagree, and how to use your rating without obsessing over it.

Online Chess Ratings Explained: Elo, Glicko and Why Your Number Differs Between Sites

Every online chess player watches a number. It goes up after wins, down after losses, and eventually becomes the main way many players judge their own progress. Yet few people know what the number actually measures, why it moves more after some games than others, or why it can be 300 points apart on two different sites.

This guide explains how chess ratings work, in plain language, and how to use yours as a tool rather than a source of stress.

A rating is a prediction, not a grade

The most important idea first: a rating is not a score of how well you play. It is a prediction of results. The difference between two players' ratings predicts how the games between them are likely to go.

The classic rule of thumb from the Elo system:

Rating differenceExpected score of the higher-rated player
050%
100about 64%
200about 76%
400about 91%

"Expected score" counts a win as 1, a draw as ½ and a loss as 0. So a player rated 200 points higher is expected to score about three points out of four — not to win every game.

After each game, your rating moves by an amount proportional to how surprising the result was. Beat someone much stronger and you gain a lot. Beat someone much weaker and you gain very little, because that result was expected. Lose to someone much weaker and you lose a lot.

Elo: the original system

The Elo system, developed by physics professor Arpad Elo and adopted by FIDE in 1970, is still used for over-the-board ratings. Its update rule is simple: your new rating equals your old rating plus a constant (called K) multiplied by the difference between your actual and expected score.

The K-factor controls how fast ratings move. FIDE uses a larger K for new and young players, so their ratings can catch up with rapid improvement, and a smaller K for established players, so their ratings stay stable.

Glicko: the system most online sites use

Online chess has a problem Elo wasn't designed for: players who play hundreds of games, then disappear for months, then come back. Most major sites therefore use Glicko or Glicko-2, systems created by statistician Mark Glickman.

Glicko adds a second number to each player: the rating deviation, a measure of how certain the system is about your rating.

  • When you are new or haven't played for a long time, your deviation is high. The system knows little, so each result moves your rating a lot.
  • As you play more games, your deviation shrinks and your rating moves less per game.

That is why your first twenty games feel wild — gains and losses of 50 points or more — and why the swings calm down later.

Provisional ratings

Many sites mark your rating as provisional (often with a question mark) while the deviation is high. A provisional rating is the system's first guess, and it can be far from your true level. If you start at 1500 but play like a 1100 player, expect your rating to fall steeply at first. That is not failure; it is the system finding where you belong.

Why your rating differs between sites

Players are often surprised that they are 1400 on one site and 1700 on another. Several reasons stack up:

  • Different starting points. Sites start new players at different numbers.
  • Different player pools. A rating only means something relative to the other players on that site. If one site's pool has more beginners, the same player will be rated higher there.
  • Different systems and parameters. Glicko versus Glicko-2, different deviation settings, rating floors.
  • Different time controls. Your blitz, rapid and classical ratings are separate, and each pool behaves differently.

So comparing ratings across sites or across time controls tells you very little. Comparing your rating on the same site and time control over months tells you a lot.

Online rating versus FIDE rating

An online rating does not convert neatly into a FIDE rating. Over-the-board play has slower time controls, longer games, different pressure and a different player pool. There are rough conversion tables floating around, but they vary widely. If you want a FIDE rating, the only way is to play FIDE-rated tournaments.

Using your rating well

A rating is useful for matchmaking — finding opponents who give you roughly even games — and for tracking long-term trends. It is less useful as a daily measure of self-worth. Some habits that help:

  • Look at monthly trends, not daily swings. A 50-point drop in a bad evening is normal statistical noise.
  • Stop after a losing streak. Rating losses from playing while frustrated ("tilt") are among the most common causes of apparent plateaus.
  • Focus on the quality of your games. If your blunders per game are going down, the rating will follow.
  • Hide your rating during games if the number distracts you. Many sites allow this.

A note on computer ratings

When you play the computer on ChessDigi, there is no rating yet — the levels are named instead: Beginner, Easy, Medium, Hard, Master. That is deliberate. The goal of those games is practice and review, not a number. When you can beat a level consistently, move up.

Try it

Rather than watching the number, watch your mistakes. Play a game against the computer, then use the review to count your blunders. Do it for a week and compare. For more on what the review labels mean, read blunder, mistake, inaccuracy.

Frequently asked questions

What is a good chess rating?

It depends on the pool. On most online sites, the median active player sits somewhere around 1000 to 1500 depending on the time control. Rather than aiming for an absolute number, aim to be in a higher percentile of players on your site than you were a few months ago.

Why did I lose so many points after one loss?

Either your rating is still provisional (high deviation), or you lost to a much lower-rated player. Both make a result more "surprising", which moves the rating further.

Is Glicko better than Elo?

For online play, it is generally a better fit because it handles players who play many games or take long breaks. Elo remains standard for official over-the-board ratings because of its long history and simplicity.

How long does it take for a rating to stabilize?

On most online sites, around 20 to 30 games in one time control. After that, each game moves the rating by a smaller amount unless you stop playing for a long time.

Need specific advice for your case?

We will contact you within 24 hours.

Request consultation now

Need advice? Talk to us

Leave your details and our team will contact you within 24 hours. The first consultation is completely free.

or
Call now 0888 831 183