ChessNextMove.ai
CalculatorAnalysisPuzzlesLeaderboardPricing
Try Pro

Chess ELO Rating Test

This chess ELO test measures your rating from results: you play rated games against Stockfish set 200 points above your current number, and the ELO formula, with FIDE's K-factors, updates it after every game. The test is part of Chess Next Move Pro.

How This Chess ELO Test Works

You begin at 1200 and play rated games against an AI opponent pitched 200 points above your current rating. Each result updates your number with the standard ELO formula and the K-factors FIDE uses for tournament play, so the rating you end up with is arrived at the way a real one would be: by results against known opposition, not by a questionnaire.

The opponent sits deliberately above you because an evenly matched game carries little information. Playing someone slightly stronger produces a more revealing spread of wins, losses and draws, which makes your rating converge faster.

The ELO Formula, and the K-Factor

Your expected score against an opponent:

E = 1 / (1 + 10^((opponent_rating - your_rating) / 400))

And the update after each game:

new_rating = old_rating + K x (actual_score - E)

actual_score:  win = 1,  draw = 0.5,  loss = 0

K decides how much one game can move you, and follows the FIDE tiers: K = 40 for your first 30 games, K = 20 once your rating is established, and K = 10 from 2400. The large early K is the point: it lets a wrong starting guess of 1200 correct itself in a handful of games instead of dozens. The rating floors at 100 and cannot go below it.

To work out a FIDE rating change for a tournament game, use the chess ELO calculator: it applies FIDE's table, K-factor and 400-point rule to a win, a draw or a loss. Chess.com and Lichess use Glicko instead, so it does not predict changes there.

How Many Games Until It's Accurate?

GamesConfidence
1–9 games±300 ELO — provisional
10–29 games±150 ELO — provisional
30–49 games±75 ELO — established
50+ games±30 ELO — accurate

One or two games tell you almost nothing. Thirty is where the number starts to be worth quoting.

Chess Rating Categories

RatingLevel
Under 800Beginner
800–999Novice
1000–1199Amateur
1200–1399Intermediate
1400–1599Club Player
1600–1799Tournament Player
1800–1999Advanced
2000–2199Expert
2200–2299Candidate Master level
2300–2399FIDE Master level
2400–2499International Master level
2500+Grandmaster level

From 2200 the levels follow FIDE's title thresholds: Candidate Master 2200, FIDE Master 2300, International Master 2400 and Grandmaster 2500. They say "level" because the real IM and GM titles also need norms earned in tournaments.

Why Your Rating Here Won't Match Chess.com or Lichess

Expect a different number, and do not assume one of them is wrong. Every rating pool is calibrated on its own population, which is why Lichess ratings generally read a few hundred points above Chess.com ratings for the same player, and why neither converts cleanly to FIDE.

What transfers between systems is not the number but the ordering: whether you are improving, and who you can hold your own against. If you want to work on that rather than measure it, the chess calculator will show you the move you should have played in any position from a game you lost, and the bot ladder runs from 400 to 3650 so you can pick an opponent just above wherever this test lands you. If you want the engine side of the same question, what ELO each Stockfish level plays at has the full table.

Frequently Asked Questions

How does the chess ELO test work?

You start at 1200 and play rated games against Stockfish set 200 points above your current rating. After each game the ELO formula, with the K-factors FIDE uses, updates your number. Because the opponent is stronger, a win is worth more than a loss costs: in your first games a win gains about 30 points and a loss costs about 10. Hints, takebacks and resets are switched off during test games.

How accurate is this chess ELO test?

It depends on how many games you have played. The test treats your rating as within about ±300 points after 1 to 9 games, ±150 after 10 to 29, ±75 after 30 to 49 and ±30 from 50 games on. It works best when your opponent falls between 1320 and 3190, the range of Stockfish's UCI_Elo setting, which is calibrated to playing strength. Stockfish cannot be set below 1320, so for players rated under about 1320 the test weakens it with a shallower search and occasional random moves, and the result is a rougher estimate.

How many games does the ELO test take?

There is no fixed number, and the more you play the better the estimate. One or two games say very little. After 10 the estimate is within about 150 points, and from 30 games, when the K-factor drops from 40 to 20 and each game moves you less, the rating is marked Established.

Is my rating here the same as FIDE, Chess.com or Lichess?

No, and it should not be expected to match. The K-factors are FIDE's, but a rating only means something inside the pool it was earned in, and this one is earned against Stockfish rather than people. The same is true between sites: one player can hold three different numbers on Chess.com, Lichess and the FIDE list. Use this rating to track your own progress rather than to compare across sites.

What is a good chess ELO rating?

Around 1200 to 1400 is a solid casual player, 1600 to 1800 an experienced club player and 2000 expert level. 2200 is where master titles begin: FIDE's Candidate Master title needs 2200, and the grandmaster title needs a rating of 2500 plus norms. What counts as good depends on where you started: gaining 200 points in a year is a real improvement at any level.

Can I retake the ELO test?

Yes. Every game keeps updating the same rating, so you never need to start over to see your progress. If you want a clean slate, the Reset button puts your rating back to 1200 and clears your game history, and the next games use the larger provisional K-factor again.

Is the chess ELO test free?

No. The rating test is part of Chess Next Move Pro. For a free reading, play the free computer opponents, from 400 up to 2600 ELO: the one you score about 50% against, counting a draw as half a point, is a fair estimate of your level, because an even score is what the ELO formula expects between equal ratings.