The engine lines
Knowing a move is a mistake isn’t enough — you need to see what should have been played. That’s the role of the Lines view of the analysis panel: the engine’s best continuations for the displayed position, with their evaluations.
Reading a line
Each line starts with its evaluation (as winning percentage or advantage), followed by the recommended sequence of moves. Three lines are shown: the best one, and two alternatives — often just as instructive, since they reveal whether the position offered several good paths or one narrow move. The ↗ icon in the header draws all three as arrows on the board (decreasing opacities) instead of only the best one.
Below the engine lines, the Continuations from here section shows the continuations of your tree — the current line always first, even without a fork, the alternatives underneath when the position offers any. The engine stays on top, in its fixed place: continuations come and go without ever pushing it around.
On faulty moves, the typical reading: line 1 is what should have been played; comparing it to the played move reveals the missed idea — a fork, a defense, a change of plan.
And the panel’s hidden gem: click a line and it unfolds as an ephemeral variation in the move list. Walk through it with ←/→, watch every position on the board — the variation evaporates as soon as you navigate elsewhere, and only gets written into the game if you play a move inside it. Deep reading that never dirties the tree.
Deepening a move
A game analysis evaluates each move once, at a set depth. Sometimes a verdict is surprising — a sacrifice classified as a Mistake, a Blunder on a move that looked natural, a tight endgame where the evaluation seems harsh. Deepen this move replays that one move to the engine at a markedly higher depth: 20, where the game analysis works at 15 and the live analysis caps at 18. That’s the first level that can genuinely say something new — recomputing at the same depth would only have mechanically confirmed what is already on screen.
Two ways to reach the command:
| Where | Gesture |
|---|---|
| The move list | Right-click the move > Deepen this move (depth 20), at the top of the menu |
| The analysis panel | The small magnifier, to the right of the current move’s card — right next to the verdict it corrects |
What gets recomputed is the designated move: the one whose classification and accuracy are displayed. The new verdict replaces the old one and is kept — in the game, in the library, in the statistics. A message sums it up: the move is reclassified if the extra depth changes its label, confirmed if it upholds it.
On the Opera Game, for instance, 15.Bxd7+ goes from 57.9% accuracy to 100%: the added depth sees the combination the game analysis had underrated.
Three usage notes. The entry only appears on a move the game analysis has already covered — not on a variation it never walked. The command is unavailable during a live game (there is no verdict to correct yet) and while an analysis is running. And the deepened result is cached at its depth: a later analysis of the game reads it back instead of redoing it, and never downgrades it.
Re-analyzing the game from scratch
The Analyze button in the toolbar, on an already-analyzed game, offers to re-analyze the game from scratch — purging the analysis cache. Useful after an engine update, or if an analysis was interrupted midway.