Chesslord

What is an inaccuracy in chess?

An inaccuracy (?!) is a move a little worse than the best one: nothing is lost on the spot, but a small part of your advantage or your chances is given away.

White’s rooks own the seventh rank. 36.Rxh5 grabs a pawn and White is still winning, but pushing the passed pawns with 36.d6 or 36.c5 was much stronger. From a game between two players rated around 1790.

An inaccuracy is the mildest of the bad labels. You didn’t hang anything and the game goes on, but there was a better move and yours cost you a little. In a game review it is marked with ?!.

What does ?! mean in chess?

In annotated games, ?! is a commentator’s mark for a dubious move. Sometimes it is a move that is objectively weak but sets a trap, or a risky attack the commentator doesn’t trust.

Game reviews on chess.com and Lichess use the same sign for something plainer: a move that was a bit worse than the best one (more on that in what best, excellent and good mean). That is why you see far more ?! in a review than in a book. Don’t mix it up with !?, which marks an interesting move. For the other end of the scale, see what makes a move brilliant.

Why is my move an inaccuracy when it looks fine?

Because it is fine, mostly. Nothing hangs and nothing is lost right away. There was simply a better move, and the review is pointing at it. Inaccuracies usually come from habits like these.

Taking the slow road in a winning position
Grabbing a pawn when a faster plan was there, like 36.Rxh5 in the game at the top of the page. You stay winning, but you give your opponent more time.
A quiet move that gives something away
A retreat or a defensive move that keeps material level but leaves a piece passive, or hands your opponent a file or a square.
An opening move that costs time
Moving the same piece twice, bringing the queen out early, or walking the king instead of castling. Nothing is lost yet, but your opponent gets a move for free.

A rook that only defends

Black is already the exchange down. 33...Rf8 guards the f6 pawn, and nothing hangs, but the rook is now stuck defending, and 34.gxf6+ just trades on f6. 33...Rh2 was best: the rook goes to the second rank and goes after White’s pawns. From a game between two players rated around 1470.

Giving up castling

After 5...c6, White plays 6.Kf1 and gives up castling for good. Nothing is lost, but the rook on h1 is stuck behind the king. 6.O-O was the simple move. In the game Black castled on the next move, and it was White’s king that stayed in the middle. From a game between two players rated around 1690.

To see what the better move was in any of your games, a game review shows it next to every move you played.

Is an inaccuracy bad?

Not on its own. The position is still there to play, and every player makes inaccuracies, strong ones included. One ?! in a game is not worth worrying about.

They start to matter in two cases: when the same one keeps showing up in your games, and when a small concession sets up a real mistake a few moves later.

Inaccuracy, mistake or blunder: what is the difference?

It comes down to how much the move gives away. An inaccuracy gives away a little. A mistake gives away a lot: a level game becomes hard to hold, or a winning one becomes level. A blunder gives away the game, or most of it.

The position from the rook example above shows all three. 33...Rf8 was the inaccuracy. Here are two other moves Black could have played.

A mistake

33...fxg5 opens the f-file for White’s rooks. 34.Rf7+ lands a rook on the seventh with check, and Black’s pawns start to fall.

A blunder

33...Rh4+ looks active, but the king simply takes the rook with 34.Kxh4. A whole rook down, the game is over.

Each site draws the lines its own way. Chess.com, for example, only calls a move a blunder when it loses material or allows a forced mate, so a move that ruins your position without losing anything concrete is a mistake there. For more on the worst end of the scale, see what counts as a blunder.

How many inaccuracies are normal?

There is no number worth chasing. It depends on the site, on the time control and on the game itself: a sharp game produces more inaccuracies than a quiet one. Counts from two different sites don’t compare either, because each one has its own scale.

Compare yourself with yourself: same site, a few months of games. If your inaccuracies are dropping over a few months and your blunders with them, you are getting better.

Why do chess.com and Lichess label the same move differently?

Because each site has its own rules and its own settings. Chess.com also has a miss label, which sometimes covers moves that would be inaccuracies elsewhere. Lichess marks only inaccuracies, mistakes and blunders, and next to each one shows the move that was best. Chesslord reviews games from both sites with the same labels.

How do you learn from your inaccuracies?

Don’t study every ?! one by one. Look for the ones that keep coming back.

Fix blunders and mistakes first
They cost far more. If your games still have pieces left hanging, start with a blunder check before every move.
Ask what the better move does that yours didn’t
Was it more active? Did it threaten something? Did it take a square away from your opponent? The answer is usually one of these.
Replay the position
Go back to the moment before the inaccuracy and try to find the better move yourself before you look.
Look for the ones that repeat
The same early queen move in every game, the same passive retreat when you are under pressure.
In a winning position, look for the forcing move
Before you grab material, check whether there is a check, a capture or a threat that finishes things faster.

Chesslord’s insights sort all your moves from brilliant to blunder across your games, so you can see whether your inaccuracies come in patterns.

How Chesslord helps

Stats from your games