Chess960 ratings: what your 960 Elo actually measures
Your Chess960 rating is not your classical rating with noise on top. It is a separate number, kept in a separate pool, and it answers a different question: how well do you play when nothing about the position is remembered.
Why are Chess960 ratings kept separate from classical ratings?
A rating is a prediction. The number exists to forecast a result when it meets another number, and the forecast only holds if both players are playing the same game against the same population. Shuffle the back rank and the game being measured is no longer the same one. A player who has spent five years on the Najdorf brings all of that to position 518 and none of it to position 34.
So the pools are kept apart. FIDE publishes its lists by time control and rates each separately. Lichess and chess.com both run Chess960 as its own pool with its own number, unconnected to your blitz or classical figure on the same account. One account can hold six or seven ratings, and no two of them are the same currency. A point won in one pool is not a point in another.
Why is my Chess960 rating lower than my classical rating?
Two reasons, and only one of them is about your play.
The first is the pool. A rating says "this strong compared to these opponents," and the opponents differ. Nobody arrives in a Chess960 pool by accident; you have to go looking for the variant and click the right button. That self-selection tends to leave a pool that is smaller and, at the top, sharper than the main one. Rated 1800 against those opponents is not the same performance as 1800 against everybody. No official conversion between the two exists. Any table you find that claims one is somebody's guess.
The second reason is what the game asks of you. Preparation is worth nothing in 959 of the 960 positions. If a real part of your classical strength is lines you know rather than moves you find, that part does not travel, and the number drops. If your strength is calculation and piece coordination, it travels intact, and plenty of players score higher in 960 than they do in standard chess.
There is a third thing that is not a reason at all but gets mistaken for one. Most online Chess960 is played at blitz and bullet speeds. If you are holding your 960 blitz rating against your classical rating, you are comparing two different clocks as well as two different games. Compare 960 blitz to standard blitz, or the comparison tells you nothing.
Where is the Chess960 rating list?
There is no single one, and any specific standing printed in an article is stale within weeks. Three places are worth checking, and all three change.
Your own platform keeps the list that actually governs your games. Lichess and chess.com each publish a Chess960 leaderboard drawn from their own pool, and your rating on one has no bearing on your rating on the other.
FIDE's site holds whatever FIDE currently publishes, along with the regulations for its Fischer Random events. Check it directly rather than trusting a number quoted elsewhere.
The Freestyle Chess circuit runs its own elite events from shuffled positions and posts its own standings. It is recent and still growing, which is exactly why it is the part of this page most likely to be out of date.
How does a smaller rating pool behave?
Rating points move between players and nowhere else. In a large pool with constant traffic, points circulate fast, and a number settles near the strength it describes. A small pool circulates slowly. A strong player who joins, wins forty games and stops playing leaves with points that never come back, and the players who fed them stay low for months. The result is more drift and a noisier number for the same underlying strength.
Small pools also mix badly. If the strong 960 players in your region play mostly each other, their ratings are internally consistent and only loosely anchored to everyone else. This is not a flaw anybody has failed to fix. It is what a rating is: a measure inside one population, not a property of you.
Why does a new Chess960 rating jump around so much?
Because the system knows it does not know you yet. Glicko-based systems, which both major sites use, track a rating deviation alongside the rating: a measure of how uncertain the number is. New account, high deviation, so every result moves your rating a long way and moves your opponent's hardly at all. Sites mark this state as provisional and usually show it with a question mark.
Around twenty games the deviation comes down and the number starts behaving. Before that, a 300-point swing in an evening means nothing. After it, a 300-point swing means something is wrong.
Is there a Chess960 world championship?
Yes. FIDE recognises the variant and awards a world championship title for it, run as its own event rather than as part of the classical cycle, and not on a fixed annual schedule. Separately, the Freestyle Chess events have grown into a circuit of invitational tournaments played from shuffled positions, with their own qualification and their own standings.
Naming a current champion here would be pointless: titles change hands, and a wrong name is worse than none. The current holders and the current calendar are on FIDE's site and on the circuit's own site.
What does a realistic Chess960 rating tell you about your play?
More than a classical rating does, in one narrow respect. Classical results are contaminated by what you happen to have memorised. Chess960 results are not, so the number is a cleaner reading of what you can see at the board.
Use it as a diagnostic rather than a score. Open your last ten 960 losses and find the move where the game turned. If the answer is regularly inside the first fifteen moves, the problem is not tactics. It is order of development, a rook stranded in a corner, or a castling right you spent without noticing what it cost. That is a fixable, mechanical problem, and the rules page covers the part of it people get wrong most often. Stepping through arrays in the position generator and asking where you would put the first three pieces is worth more than another blitz game.
If the losses cluster after move twenty instead, the openings are fine and you are missing things. Pattern recognition on unfamiliar squares is a separate skill from pattern recognition on familiar ones, and it drills the same way any other tactic does. The Chess960 puzzle trainer uses positions from real games with engine-checked solutions, and the books hold a thousand of them at a time, sorted by rating so you can start where your number actually is.
One last thing about that number. It is a measure of results in one pool at one time control, nothing more. It is not a rank among all players, it is not transferable, and it is not you. Play enough games for the deviation to settle, then ignore it and go work on the losses. If you want the vocabulary that comes with the variant, the Chess960 glossary defines the rest.