Free chess game review
Unlimited, no account, no ads: see every brilliant move and blunder, and the reason behind each.
Or try a famous game
How it works
A review in three steps
1
Pick a game
Type a Lichess or Chess.com username, paste a game link, or paste a PGN.
2
Peck analyses it
Stockfish runs on your device. The graph fills in at once and the labels settle in seconds.
3
Review and train
Step through the key moments, retry the better move, and turn your mistakes into puzzles.
Open method
How Peck labels moves and computes accuracy
Every label is a measured loss of winning chances, checked on the board. Here is exactly how.
Brilliant. A sound sacrifice: it leaves a piece the opponent can win, and is still the best move or nearly.
Great. The only good move: the next best is clearly worse.
Best. The engine's top choice.
Excellent. Within 2% of the best move, in winning chances.
Good. Within 5% of the best move.
Theory. A known opening move, from Lichess's open list of named openings.
Forced. The only legal move.
Inaccuracy. Gives away up to 10% of your winning chances.
Mistake. Gives away up to 20%, or more without losing material or allowing mate.
Missed win. Lets a chance slip after your opponent's error, or gives up a forced mate.
Blunder. Gives away more than 20% and something concrete: material, or a mate.
Accuracy and “played like”
- Winning chances follow a curve that depends on rating, as on Chess.com: a stronger player converts a smaller edge, so the same slip costs them more.
- Each move's accuracy comes from the winning chances it kept: 103.17 × e^(−15.5 × loss) − 2.17, from 0 to 100. Theory counts as perfect; forced moves don't count.
- A game blends the average with the harmonic mean (88.4% and 11.6%), so one blunder weighs more than a few small slips.
- "Played like" moves your rating by how far your accuracy sits from what players at your rating usually reach, and is always shown as a range.
FAQ
Questions about game review
What is a brilliant move?
A brilliant move gives up material on purpose, leaving a piece the opponent can take, and is still the best or nearly the best move. Peck checks the sacrifice on the board itself and re-checks every brilliant with a deeper search before it shows the label.
Is it really unlimited?
Yes. The engine runs in your browser, so a review costs Peck nothing to run. There's no account, no daily cap and no ads. A free account only adds share links and turning your mistakes into puzzles.
Does it work with Chess.com?
Yes. Type your Chess.com username to list your recent games, paste a game link (Peck asks for your username, because Chess.com only shares games through the players' archives), or paste the PGN from the game's share menu.
Where does the analysis run?
On your device. Peck sends Stockfish 19 to your browser and analyses the game there, so nothing is uploaded. Finished reviews are kept in your browser, so opening one again is instant.
Why do labels differ from Chess.com sometimes?
Engines disagree: about a quarter of labels change between engine builds alone, and Chess.com's quick summary and full review disagree with each other too. Peck follows Chess.com's published, rating-aware model, re-checks Brilliant, Great and Missed win with a deeper search, and never changes a label once it is shown.
Beyond one game
Train the mistakes from all your games.
Peck finds the tactics you missed in your recent Lichess or Chess.com games and builds a Woodpecker set from them.
- 20–500 puzzles
- Choose your cycles
- Theme and rating filters
- Always free
