Quick answer
Quick answer: You almost certainly can't tell, and neither can most titled players. Accuracy scores don't prove it, fast moves don't prove it, and a single crushing game proves nothing at all. Report it and let the detection systems decide.
That's an unsatisfying answer and it's the correct one.
I read what both major sites publish about their fair play work before writing this. The problem itself has a long history, described on the cheating in chess page, and the modern version is a statistics problem rather than an eyeball one.
Why your instinct is usually wrong
You notice the games you lose badly. You don't notice the ones where your opponent played fine and you played worse.
That selection is the whole illusion. A player who beats you cleanly looks suspicious precisely because the loss stung.
And the games you win against somebody having a bad day never trigger the same thought.
So the feeling is real and the evidence behind it is one game, remembered from one side.
What accuracy scores actually mean
A high accuracy figure is the most misread number in online chess.
Accuracy compares your moves against an engine's preferences in the position you were actually in. Simple positions produce high accuracy for everyone, because there's little to get wrong.
A quiet game where pieces come off early can hand a club player a very high score with no help at all.
Our explainer on what an evaluation score means covers why that number moves the way it does.
Fast moves are weak evidence
People assume a cheater plays slowly because they're consulting something. Often the opposite gets reported.
Familiar positions get played instantly by strong and weak players alike. Opening theory, standard recaptures, forced sequences. None of that needs thought.
Meanwhile a real engine user often moves at a natural pace on purpose.
So speed tells you about familiarity, not about honesty, and it points in both directions at once.
What the detection systems look at
They don't look at one game. That's the key difference between their method and yours.
Both major platforms publish a description of their approach, and the shape is the same. They compare a player's move choices against expected behavior across many games, then weigh how unlikely the pattern is.
One brilliant move is noise. A sustained pattern across hundreds of positions is signal.
You can read the outlines on the chess.com fair play page and the lichess fair play page.
The one thing that is genuinely suspicious
Consistency at the wrong level, sustained over time.
A player who finds long forcing sequences repeatedly, avoids every trap, and never has an off day is doing something a human rating band doesn't produce.
Humans are uneven. Even strong players blunder, get tired and misplay endgames.
Smooth, level, error free play across many games is the pattern the systems are built to catch, and it's not something you can judge from one match.
Rating is not proof either way
A 1200 rated account playing like a 2000 could be a cheater. It could also be a returning player, a new account from a strong club player, or somebody who just improved.
Sandbagging and smurfing look identical to cheating from the losing side of the board.
So a mismatch tells you the rating is wrong, which is a different problem with different fixes.
Our note on why ratings differ between platforms covers how far apart the same player's numbers can sit.
What to do instead of guessing
Report it and move on. That's not a shrug, it's the mechanism working as designed.
A report adds your game to a queue where it gets examined against a much larger picture than you have access to. Your single data point is genuinely useful there.
What isn't useful is accusing them in the chat, which achieves nothing and often costs you your own account standing.
Then close the tab and take a break, because tilt costs you more rating than any cheater will.
Do not expect to be told
Neither platform tells you the outcome of your report, and people find this maddening.
The reason is straightforward. Publishing how a case was decided teaches the next cheater exactly what to change.
So the silence is deliberate and it's the right call, even though it feels like nothing happened.
Sometimes you'll see a closed account weeks later. Often you'll never know, and that has to be acceptable.
Review the game properly first
Before you report, look at what actually happened. Half the time you'll find the answer there.
Run the game through an analysis board and look at your own moves rather than theirs. Where did the position slip? Was it one blunder or a slow drift?
That's the review that makes you better, and it usually kills the suspicion at the same time.
Our guide on analyzing your games effectively covers what to look for and in what order.
The opening trap explanation
A lot of suspicion comes from losing quickly to an opening line you didn't know.
That isn't an engine. That's preparation, and it's the most common thing in online chess.
Somebody who plays the same line five hundred times knows every trap in it, and it looks like magic from the other side the first time.
The cure is learning the line rather than reporting the person.
Chat behavior tells you nothing
A rude opponent feels more likely to be cheating. That's a feeling about manners, not about moves.
Plenty of unpleasant players are entirely honest, and plenty of cheaters say nothing at all because silence draws less attention.
The same goes for a brand new account with a default picture. Everybody starts somewhere.
So strip out how the person behaved and look only at the moves, which is exactly what the detection systems do.
Disconnects and long thinks are normal
A pause on move twelve gets read as somebody switching windows. Usually it isn't.
People get up, answer the door, lose their connection, or genuinely think about a critical position. All of that looks identical from your side of the board.
A long think before a hard decision is a good sign about your opponent, not a bad one.
What matters is the pattern across a game, and even then a single game can't carry that weight.
Where cheating actually concentrates
It clusters in specific places, and knowing them makes your judgement better.
Prize events, rating milestones and long time controls carry more of it, because there's something to gain and time to use a second screen.
Bullet and hyperbullet carry much less of it, since consulting anything costs more time than it saves.
So a suspicious loss in a one minute game is almost always just a fast opponent.
Playing bots removes the question entirely
If the suspicion is spoiling the game for you, there's a clean alternative.
A bot at a set strength plays the same way every time and there's nothing to wonder about. You can work on a weakness without the emotional noise.
It's also better practice than tilted games against strangers, because you control the difficulty.
You can set the level and play on our board whenever the online pool stops being fun.
Where the odds actually sit
Cheating exists and both platforms close a large number of accounts for it. That part is real.
What's also real is that most individual accusations are wrong, and the person accusing has one game and a bad feeling.
Holding both of those at once is the mature position, and it's the one that keeps you improving.
I compared a set of my own losses against my first impressions and the suspicion held up in none of them. Every one was my own mistake, found in the review.