Free play and diverging from a game

Move the pieces yourself and have every move scored as you make it.

Reviewing a finished game tells you what went wrong. It is less good at fixing it, because you already know the answer by the time you look. Free play inverts that: you make a move without knowing the verdict, and the engine responds immediately. The app opens in this mode, on the standard starting position, so there is nothing to set up.

What happens when you move

Every move you play is analyzed and graded on the spot, with the same labels used for imported games — see how moves are classified. Alongside the grade you get:

The Best chip: undo and replay in one tap

When your move was not the engine's choice, the feedback row shows a tappable chip naming the better move. Tapping it takes your move back and plays the engine's instead, leaving you in the position you would have reached had you found it.

This is the feature worth building a habit around, because it closes the loop immediately. Instead of noting "I should have played Nd5" and moving on, you are standing in the position that move creates, and you can carry on from there to see why it was better. The mistake and its correction stay in the same session.

A small correctness detail, since it is the kind of thing that goes wrong quietly elsewhere: the suggested move is always recalculated against the position actually on the board. A cached engine result can describe a different turn — a different player's move even — and displaying it verbatim would suggest something illegal or nonsensical. The name you see always belongs to the position you are looking at.

The controls

ControlDoes
UndoTakes back the last move, one at a time
ResetReturns to the position the session started from — the standard start, or the position you imported
FlipTurns the board around; the evaluation bar follows
Copy PGNThe moves played this session, to the clipboard
Copy FENThe current position, to the clipboard

Reset returning to the imported position rather than the opening is what makes drilling practical: set up a position once, then reset back to it as often as you like without re-importing.

Diverging from an imported game

The same machinery works part-way through a game you have loaded. Step to the moment that interests you, play a different move, and you are in your own line with everything still scored. The engine has no opinion about whether you are "following the game" — it evaluates the position in front of it.

That is the honest answer to the question every post-mortem raises: what if I had played the other move? You can find out, several moves deep, instead of guessing.

Drilling a position you keep getting wrong

  1. Import the position as a FEN, or step to it in a game — see importing a game or a position.
  2. Play the move you would play under time pressure.
  3. If it is graded badly, tap the Best chip and study the position the better move produces.
  4. Reset and play it again, now knowing what you missed.

The value is in step two: committing to a move before seeing the verdict. Reading an annotation is not the same skill as choosing under uncertainty, and only one of them is what you do in a game.

Worth knowing: free play always analyzes at the default depth, regardless of the depth chosen in Settings — that choice applies to analyzing an imported game. If you want a position examined more deeply, import it. See choosing an analysis depth.

Start playing →