Chess ratings explained, from 600 to 1800

What a chess rating measures

A rating sums up your past results in one number, and that number helps predict the next ones. When two players of the same pool meet, the gap between their ratings says how much the stronger one is expected to score. Score more than expected and your rating goes up; score less and it goes down.

The method comes from Arpad Elo, a physics professor and a strong chess player. FIDE, the international federation, adopted his system in 1970 and still uses it for games played over the board. Its regulations include a table that turns a rating gap into an expected score: a player rated 200 points above their opponent is expected to score 0.76 points a game, about three points out of four.

Elo, Glicko and Glicko-2

Lichess and Chess.com use methods developed by Mark Glickman: Glicko-2 on Lichess, Glicko on Chess.com. Beside your rating, both keep a second number, the rating deviation, which says how sure the system is about it.

That is why a new player’s rating swings a lot in the first games, and why the steps shrink as the system gets to know you. On Lichess, a question mark next to a rating means it is still provisional.

Why Lichess, Chess.com and FIDE ratings differ

A rating means something only inside its own pool of players. The Lichess FAQ advises against comparing ratings between organisations: their players are not of the same level, their rating systems differ, and both can change the numbers a great deal. They do not even start from the same place:

• Lichess starts every new player at 1500, with a margin of uncertainty of 1000 points either way.
• Chess.com asks new members to choose a skill level, which sets their starting rating: 400 or 800 for someone new to the game, higher for a player who says they are stronger.
• FIDE publishes a first rating after at least five games against rated opponents, and since March 2024 that rating must be at least 1400: a player whose rating drops below 1400 is shown as unrated on the next list.

So the same player can carry quite different numbers on two sites, and a rating of 1000 cannot be a FIDE rating today: it is an online rating, or one from a national federation. Time controls add more numbers. Lichess and Chess.com keep a separate rating for each time control, such as blitz and rapid, and FIDE has separate ratings for standard, rapid and blitz. Compare yourself with the same rating, on the same site, over time.

Puzzle ratings and game ratings are different scales

Puzzles have ratings of their own. On Lichess, each attempt at a puzzle counts as a rated game between you and the puzzle, so every puzzle has a rating, and so do you, separately from your game ratings. Chess.com rebuilt its puzzle ratings in October 2025 and re-rated every player and puzzle, because the old system had inflated them.

The two numbers measure different things. In a puzzle you know there is something to find, you can usually take your time, and there is one position to solve. In a game nobody warns you, the clock runs, and you have to keep finding good moves until the end.

A puzzle rating says how well you find tactics when you look for them; a game rating also counts openings, plans, endgames and the clock. That is why a rising puzzle rating does not promise a rising game rating, and why we never promise one either.

What players usually know at each level

No number tells you exactly what a player knows, and two players with the same rating have different strengths. Read what follows as tendencies, not rules. In everyday speech people say Elo for any chess rating, online ones included, and these three bands are also the levels of our notebooks.

600 to 1000: the moves are known, pieces still hang

Players here know how the pieces move and play whole games. Games are often decided by a piece left where it can be taken, or a mate in one missed by both sides. Patterns such as the fork, the pin and the back-rank mate may be known by name, but they are not yet seen at a glance. A useful habit: after each of your opponent’s moves, look at what it attacks.

1000 to 1400: tactics found when you look for them

Players in this band leave fewer pieces hanging, and they often solve a fork or a pin when they know one is there, as in a puzzle. In their own games the same tactics slip by, because nothing announces them. Games tend to turn on a two-move tactic, a defence overlooked or an endgame that should have been won. A useful habit: before each move, list the checks, captures and threats for both sides.

1400 to 1800: calculation and plans

Here most simple tactics are seen, and players calculate a few moves ahead. What is often missing is the quiet move that prepares a combination, the defence that almost holds, or a plan when there is no tactic at all. Openings and endgames start to matter more. A useful habit: when you find a good move, look for a better one, and check your opponent’s best reply before you play it.

How to use your rating to train

Use your rating to choose your material, not to judge yourself. Pick puzzles and books for the band you are in, move up when they start to feel easy, and watch the trend over months on one site and one time control rather than the swings of a single evening. If that trend has stayed flat for months, our article on what to do when you are stuck suggests what to change.

The Knight’s Notebook follows the three bands of this article: Beginner (600 to 1000 Elo), Intermediate (1000 to 1400) and Advanced (1400 to 1800), with sixty puzzles a month and every solution explained. If you play online, your rating there is a good guide to the band; numbers differ between sites, so if the first notebook feels too easy or too hard, change level. 300 Chess Puzzles covers the whole range in one book, from about 600 to 1800 Elo, easiest first in each theme. Our article on how many puzzles to solve a day turns this into a routine, and the one on choosing an opening shows what club players really play.