How to analyze your game with a free engine

A free analysis engine is more than enough to improve. Here's how to actually use one, without falling into its traps.

Key takeaway

Replay your game without an engine first, to form your own judgment, then use the engine to check it: it's the combination of both that actually drives progress, never one without the other.

87654321

abcdefgh

Here's the kind of moment where the evaluation swings suddenly: Black's bishop on c3 is hanging (bxc3 wins it outright). This is exactly the kind of sharp swing an engine spots instantly, where an untrained eye can easily miss it.

Another article on this site covers the manual method for analyzing your own game, without any tool at all. This one is its practical companion: how to actually equip yourself with a free analysis engine, and above all how to use it without falling into its traps.

No need to pay

A free analysis engine (the one built into Lichess, for instance, entirely free and requiring no account for occasional use) is more than enough at this level. Paid engines gain precision on extremely fine calculations, a level of detail that brings nothing concrete to an improving player.

Importing your game

Two main methods exist for loading a game into an analysis tool: pasting the PGN directly (the game's text, in standard move-by-move notation) into the dedicated import field, or, if the game was played online, simply grabbing its ID or link to open it directly.

Reading the evaluation graph

Most tools display a graph or evaluation bar that tilts one way or the other as the game unfolds. A number like +1.5 means, in plain terms, that White has an advantage roughly comparable to a pawn and a half; a number like -3 indicates a Black advantage worth roughly three pawns. The article "understanding a chess engine's evaluation" covers this mechanism in more depth.

The vocabulary of move classifications

✓Excellent: an optimal move, or nearly so, often the only one that keeps the advantage
✓Good: a solid move, without giving up any meaningful advantage
✓Inaccuracy: a move that gives up a bit of ground without being outright bad
✓Mistake: a move that gives up a real advantage, without being catastrophic
✓Blunder: an error that radically flips the evaluation, sometimes losing the game outright

Look at the diagram

87654321

abcdefgh

Black's bishop on c3 is hanging: bxc3 wins it immediately. This is exactly the kind of sudden evaluation swing an engine detects in a fraction of a second.

The classic trap: following the engine without understanding

The biggest mistake is blindly accepting the engine's "best move" without trying to understand the human idea behind it. An engine gives a move, rarely an explanation; it may recommend a move for reasons far too deep or abstract to be worth memorizing. What matters is understanding why a move was better, not memorizing it.

A concrete working method, in five steps

✓1. Replay the whole game first without an engine, noting your own impressions of the moments that feel critical
✓2. Then run the engine to check or correct those impressions
✓3. Focus on the moments where your judgment and the engine's diverge the most
✓4. Try to understand the idea behind the best move, not just observe it
✓5. Keep only one or two clear takeaways per game, rather than ten vague remarks

Key takeaways

✓A free engine is more than enough to improve at this level
✓The evaluation graph reads as a "pawn-equivalent" advantage for one side or the other
✓Classifications run from excellent to blunder, passing through inaccuracy and mistake
✓The engine gives a move, not always a reason: trying to understand still matters most
✓Replay without the engine first, check with it afterward, and keep only one clear takeaway

Check your understanding

In engine-analysis vocabulary, what's a mistake called that immediately loses the game or a decisive amount of material?