The analysis bar shows a number like +1.5, -2.0, or #3, and it tells you who is winning and by how much. Once you can read it, every engine line and evaluation graph starts to make sense. This guide explains what the number means for you as a player — not how the engine computes it, which is a separate topic.
The Sign: Who Is Better
The evaluation is always given from White’s point of view. The sign tells you the direction of the advantage:
- Positive (+) — White is better.
- Negative (−) — Black is better.
- 0.0 — the position is balanced.
So +1.2 and −1.2 describe equal-sized advantages for opposite sides. A score of −2.0 does not mean “bad”; it means Black is winning by about two pawns' worth. If you played the black pieces, −2.0 is great news.
The Number: Pawns and Centipawns
The value is measured in pawns. +1.0 means “about one pawn’s worth of advantage” — though that advantage is often positional (more space, a safer king, the bishop pair) rather than a literal extra pawn.
Internally engines work in centipawns, where one pawn equals 100 centipawns. A raw score of 150 cp is shown to you as +1.5. The decimal just lets the engine express small edges precisely.
A score around +1.5 is a clear, real advantage — but not yet a forced win.
Rough Thresholds You Can Trust
You do not need to memorise exact cutoffs, but these bands are a reliable mental map:
| Eval | Meaning |
|---|---|
| 0.0 to ±0.5 | Roughly equal — any result is possible |
| ±0.5 to ±1.5 | A small but real edge; the better side presses |
| ±1.5 to ±3.0 | A clear advantage; should win with accurate play |
| beyond ±3.0 | Usually winning — often a piece or more ahead |
| #N (mate) | Forced checkmate, result decided |
Let us see what these look like on the board.
Diagram 1: Roughly Equal (≈ 0.0)
Most openings produce a near-zero evaluation. Neither side has done anything wrong yet, so the engine sees a balanced fight.
A tiny positive number this early just reflects White’s first-move initiative. Treat anything inside ±0.5 as “the game is still wide open.”
Diagram 2: A Clear Edge (≈ +1.5)
Here White is a clean pawn up in a knight endgame. There is no immediate knockout, but the extra pawn is permanent and meaningful.
At +1.5 the better side is genuinely on top and should aim to win, but one careless move can still let it slip back toward equality. This is exactly the kind of position where reviewing your own technique pays off.
Diagram 3: Mate in 3 (#3)
When the engine finds a forced checkmate, it stops reporting pawns and shows a mate score instead — #3, M3, or +M3. The exact number is irrelevant; the game is already decided.
A score like −#2 is the same idea from the other side: the opponent has a forced mate against you in two. Mate scores always trump pawn scores — there is nothing better than mate and nothing worse than getting mated.
The Number Assumes Best Play
This is the single most important caveat. Every evaluation answers one question: *if both sides play perfectly from here, what is the result? * It is not a prediction of what will actually happen in your game.
That is why a position can read +0.3 even though a tactic is on the board — the engine assumes you will find the best defence. The moment a human misses it, the real, practical evaluation is very different.
Why the Eval Swings
Because the score assumes best play, it reacts sharply to imperfect moves:
- A small inaccuracy might nudge it by 0.3–0.5.
- A mistake can move it a full point or more.
- A blunder can flip it from 0.0 to −4.0 in one move.
Those jumps in the evaluation graph are exactly where your mistakes live. Each big step up or down marks a move that changed the verdict — and that is the move worth studying. This connects directly to move accuracy and labels: a blunder is just a move that cost a large chunk of evaluation.
Reading the Number in Your Own Games
When you review a game, walk the evaluation graph and watch for cliffs:
- Find where the line drops (your mistakes) or jumps (your opponent’s).
- Click that move and compare it with the engine’s best move.
- Ask *why* the score changed — a hanging piece, a missed tactic, a king left exposed.
That habit turns an abstract number into concrete lessons. For a full walkthrough, see how to analyze your games.
Get Started
Import your games from Chess.com, Lichess, or upload PGN files
Quick Reference
- Sign = who is better (+ White, − Black).
- Number = how much, in pawns (±0.5 equalish, ±1.5 clear edge, ±3 winning).
- #N = forced mate; result decided.
- Always assumes best play — that is why it swings when a real move is imperfect.
Read the bar, find the swings, learn from the moves that caused them.
Related
- Understanding Chess Accuracy — how move labels and accuracy % are derived from evaluation.
- How to Analyze Your Chess Games — a step-by-step review routine.
- How Stockfish Evaluates Positions — the engine internals behind the number.
- Free Chess Analysis — run the evaluation on your own games.