You’ve just finished a game on Chess.com and something felt wrong.
Maybe your opponent played an almost flawless game. Maybe they suddenly found a series of strong moves in a complicated position. Or perhaps their Chess.com accuracy score came back at 98%.
Were they cheating?
Unfortunately, there’s no single statistic that can answer that question. A strong game can be completely legitimate, and even a very high agreement with Stockfish doesn’t prove that someone used an engine.
What you can do is look for statistical evidence that the moves played were unusually engine-like.
Can you tell if someone is cheating on Chess.com?
You generally cannot determine with certainty that someone cheated from a single chess game.
Chess.com's own Fair Play system considers many different signals when investigating players. Its publicly described methods include comparing players' moves with computer moves and assessing the statistical significance of those similarities.
Finding the engine's top move isn't unusual. Humans find the best move all the time, particularly when the move is obvious, forced, or part of a well-known opening.
How often did this player find engine-quality moves in situations where a player of their strength would not normally be expected to find them? is a harder question and the basic idea behind statistical chess cheat detection.
What are the signs that a Chess.com game might be suspicious?
No individual signal proves engine assistance. Several signals together, however, can make a game statistically unusual.
1. High engine agreement
One obvious place to start is to compare the player's moves with those recommended by a strong chess engine such as Stockfish.
If a player repeatedly chooses the engine's first choice, that may look suspicious.
But raw engine agreement has a major problem: sometimes the best move is easy to find.
Consider a position where:
- there is only one legal move
- a queen can be captured for free
- an obvious checkmate is available
- the position is still following established opening theory
A human player and Stockfish will often choose the same move in these situations.
Engine agreement becomes more interesting when the position gives the player many plausible alternatives and the engine's preferred continuation is difficult to discover.
2. Very low centipawn loss
Centipawn loss measures how much a player's move worsens their position compared with the engine's preferred move.
A player with a very low average centipawn loss has played an extremely accurate game.
A simple game may naturally produce a low centipawn loss. If your opponent gets a winning position early and then converts it using obvious moves, their final statistics can look extraordinary without anything unusual having happened.
Low centipawn loss is therefore useful evidence, but it shouldn't be considered in isolation.
3. Exceptional moves in difficult positions
This is one of the most useful distinctions.
Imagine two positions:
Almost every reasonable move keeps roughly the same evaluation.
There are 30 legal moves, several look attractive, and only one difficult move preserves the advantage.
Matching a chess engine in the second position is much more informative.
This is why position complexity matters when analyzing a chess game for possible engine assistance. A useful detection system should distinguish between moves that many humans would naturally find and moves that are much less likely to be discovered without deep calculation.
Read more: How to Measure Chess Position Complexity
4. Performance far above what you would expect from the player's rating
A 1000-rated player can absolutely play an excellent game.
The important question is whether their performance is unusual relative to what we'd normally expect from a 1000-rated player.
For example, finding one difficult Stockfish move isn't particularly surprising.
Finding a long sequence of difficult, highly accurate moves may be much less likely.
A statistical detector can therefore compare the observed quality of a player's moves against the expected performance of a human player of similar strength.
5. A repeated pattern across multiple games
Evidence becomes much stronger when an unusual pattern repeats.
Any player can have the game of their life. Strong tactical sequences happen. Opponents blunder. Preparation can produce nearly perfect openings.
If the same player repeatedly produces statistically exceptional engine agreement in difficult positions, the probability that you're simply observing an unusually good game becomes smaller.
This is one reason Chess.com's own Fair Play analysis isn't based solely on the Accuracy score from one game.
Does 95% or 99% Chess.com accuracy mean someone cheated?
No.
A very high Chess.com accuracy score is probably the most common reason players become suspicious of an opponent, but accuracy alone isn't evidence of cheating.
A short game can easily produce extremely high accuracy.
So can:
- a game containing many forced moves
- a heavily theoretical opening
- an opponent making serious mistakes
- a simple winning conversion
- somebody simply playing an excellent game.
A 99% accuracy game can be legitimate. A cheater also doesn't necessarily need to achieve 99% accuracy.
The important information is hidden beneath the headline percentage: which moves were played, how difficult those moves were, and how surprising that performance was for the player involved.
Read more: Does High Chess Accuracy Mean Someone Is Cheating?
Can Stockfish detect chess cheating?
Stockfish can tell you which moves are strongest.
It cannot, by itself, tell you whether the person playing those moves cheated.
If Stockfish's top move is Qxd5, knowing that your opponent also played Qxd5 tells you very little without knowing the position.
Perhaps Qxd5 wins a free queen and virtually everyone would play it.
Alternatively, perhaps Qxd5 is a positional sacrifice that only becomes best after a long engine calculation.
A chess cheat detector therefore needs to do more than simply count Stockfish matches.
How does Chess.com detect cheating?
Chess.com doesn't publish the complete details of its Fair Play system, for the obvious reason that doing so could make it easier to evade.
Its public documentation says that its analysis includes comparing human moves with computer moves and assessing their statistical significance.
Chess.com also says its broader Fair Play system uses automated analysis alongside human review and evaluates many factors across a player's games.
That means the Accuracy number shown after a game should not be confused with Chess.com's cheat-detection system as they serve different purposes.
Chess.com Accuracy is intended to describe the quality of the chess played.
Fair Play analysis is trying to answer a much more difficult question: whether that pattern of play provides evidence of prohibited outside assistance.
What counts as cheating on Chess.com?
In live games, Chess.com prohibits outside assistance.
That includes using chess engines such as Stockfish to analyze the position while the game is in progress, as well as receiving assistance from another person.
There are some differences in the resources permitted in Daily Chess, so it's best to check Chess.com's current Fair Play rules if you're unsure.
Analyzing a game after it has finished is different. That's exactly what post-game engine analysis and tools such as Chess Cheat Detector are designed to do.
How can I analyze a Chess.com game for cheating?
You can analyze a finished Chess.com game with ChessCheatDetector.com.
Paste the URL of the Chess.com game and the site will retrieve the game and analyze the moves using Stockfish alongside statistical measures designed to identify unusually engine-like play.
Rather than treating every engine match equally, the analysis considers information such as:
- engine agreement
- centipawn loss
- the player's rating
- expected human performance
- the complexity of the positions in which strong moves were found.
The result is an assessment of whether the moves played were statistically unusual compared with what we'd expect from human play.
Don't accuse someone based on one unusual game
Chess cheating detection is inherently probabilistic.
That's why a responsible analysis should distinguish between “this player cheated” and “this game contains statistically unusual signs of engine-like play.”
If you believe someone actually violated Chess.com's Fair Play rules, report the account through Chess.com and allow its Fair Play team to investigate.
If you're simply curious about a game that felt suspicious, you can analyze it and see what the moves themselves suggest.