Chess Engine Evaluation Explained
What +0.2, +1.5, +5, and M3 actually tell you.
Six positions, six evaluations
Nearly equal
+0.2The starting position. Both sides have identical material and no weaknesses yet — the small +0.2 just reflects White's first-move advantage, nothing more.
Modest advantage
+0.8Black is simply missing the e7 pawn, with no compensation. An extra pawn with nothing else going on is a real but modest edge — enough to matter over a long game, far from decisive on its own.
Winning advantage
+3.7Black is down a full rook for nothing. +3.7 is a clearly winning material advantage — from here White should be able to convert with normal, careful play.
Forced mate
M1Ra8 is checkmate — the black king is boxed in by its own pawns with no escape square. Once mate is forced, the engine reports it as "M1" instead of a centipawn number: no amount of material matters once the king can't escape.
Before: material equal
0.0Queen and king each, nothing else on the board. Dead equal — 0.00.
After: White's queen is gone
-M12Same position, minus White's queen — as if it had just been blundered away. The evaluation doesn't just drop, it flips to a forced mate against White: king and queen force checkmate against a lone king in every line, and the engine can already see the whole way there.
Want to see the evaluation update live as you make moves?
A chess engine's evaluation is a single number that summarizes who's better and by roughly how much, always from the perspective of the side to move — positive favors them, negative favors the opponent. It's measured in pawns: +1.00 is worth about one pawn's advantage.
It is not the same thing as material. An engine weighs king safety, piece activity, pawn structure, and concrete tactics alongside material — which is exactly why the evaluation can swing sharply after a single move even when no pieces are captured.
Reading the scale
Under about ±0.5, the position is close to equal — the kind of edge that exists in most normal opening positions. From roughly ±1 to ±2, that's a real, meaningful advantage, comparable to being up a clean pawn. Past +3 or so, it's generally winning with accurate play. None of these are hard cutoffs — they're a feel for the scale, not a guarantee of the result.
Centipawns
Internally, engines compute in centipawns — hundredths of a pawn — because whole pawns aren't precise enough to rank close alternatives. +218 centipawns is the same thing as +2.18; sites and apps almost always convert this to the more readable decimal form automatically.
Mate scores
Once an engine has calculated a forced checkmate, the evaluation switches from a centipawn number to a mate score like M3 (mate in 3 moves) or -M4 (getting mated in 4). At that point the centipawn scale stops mattering entirely — no amount of material saves a position with a forced mate against it.
Why evaluation can swing after one move
Material and evaluation aren't the same because a position's evaluation also depends on things a raw piece count can't capture — an exposed king, an undefended piece, a piece that's about to be trapped. A single move can create (or lose) a decisive tactic without a single piece leaving the board, which is exactly why the score can jump by several pawns' worth in an instant.
Depth and why engines can disagree
Depth is how far ahead the engine has searched along its main lines. A shallower search can miss a tactic that only becomes clear several moves deeper, which is why two engines — or the same engine at two different depths or time limits — can report meaningfully different evaluations for the exact same position. Given enough depth, strong engines converge; in practice, everyone is working with a time-bounded approximation.
Frequently asked questions
What does a positive vs. negative evaluation mean?
It's always from the side-to-move's perspective: positive means they're better, negative means their opponent is. It is not fixed to White or Black.
What's the difference between +0.2, +1.5, and +5?
+0.2 is close to equal. +1.5 is a real, meaningful edge — roughly like being up a clean pawn. +5 is a decisive, usually winning material or positional advantage.
Why do two engines sometimes disagree about the same position?
Different engines (or different search depths/time limits on the same engine) can reach different conclusions, especially in sharp or unclear positions — deeper search generally finds the truth more reliably, but every practical evaluation is a time-bounded approximation.
Is it free?
Yes. The board is read and analyzed right in your browser, there is no sign up, and you can do it as many times as you like.