Detecting Cheaters

I have reported a user for suspected cheating twice now.

Even though that person has not been punished yet, I want to know the opinion of people from this community, whether they agree or disagree with my suspicion.

There are several key features / signs that can tell whether a player is cheating or not:

* A user takes approximately the same time for every move (a clear indicator for a cheater in cases when the move is easy or forced) - the reason behind that is cheaters need to switch to the engine and enter the next move, which usually takes around 3-5 seconds
* A user makes unnatural moves; a move that a human typically would not make - the reason behind that is that those moves avoid mistakes/inaccuracies according to the engine, calculating many moves ahead
* A user has a very high success rate and thus the ELO progress jumps rapidly towards IM/GM level

Now I present you a rapid progress of this user:



And here are the move times of some of his games (left column: as black, right column: as white):



In my opinion, the images clearly show the progress & games of a cheater:

* a very rapid progress, after he chose to cheat because he just could not get above ~1200 score for several weeks - or maybe he learned how to play so much better within a very short time frame (lets say 1 week)?
* very similar move times patterns: At first, very natural move times, but then he takes a very long time (1 minute or more at blitz) to set up the game in an engine, then uses the same amount of time for every move. It does not matter what type of moves they are: very complex or very easy/forced moves: all of them take the same amount of time.

So, am I correct to assume this guy is a cheater, or are you of different opinion?

Note that most of the games look like this, these are NOT a manual selection by searching the "correct" ones out of 100s of games.

rather see specific games which u think cheating and analyze them see their centipawn loss mistakes and all things if centipawn loss below 10 no mistake or 1 or 2 inaccuracy then report if lichess detects he used engine they will ban .

Eh, this really isn't that suspicious. Rating gain usually comes in a run of wins due to improvement in your playing. While gaining 400 points over a month is certainly impressive, its not completely unheard of at this level. If one fixes a core issue or many core issues in their playing possibly through the help of a good mentor or excessive time spent playing, then I wouldn't think it to be absurd for their rating to increase dramatically.

The move time consistency could be due to cation on easy/forced moves or simply that they are using a trackpad and it takes a while to enter a move. In fact, you could have told me that he played the white pieces on the left column's images and I wouldn't be any more bemused.

I would need to see the moves themselves to judge whether they are truly unnatural for I, no offense intended, simply don't trust the judgement of a 1300 on moves being unnatural.

1600 is not GM/IM level, 2000 is not GM/IM level, 2200 is not GM/IM level, 2500 lichess is the very least one could possibly consider to be the very lower end of a titled player's rating.

@plsdontclosemyacc said in #2:
> rather see specific games which u think cheating and analyze them see their centipawn loss mistakes and all things if centipawn loss below 10 no mistake or 1 or 2 inaccuracy then report if lichess detects he used engine they will ban .

My suspicion though is that this person first does some (inaccurate) moves, then activates the engine. Or throws in some moves of his own from time to time. So there are some inaccurate moves or maybe blunders in the games, so its not 100% accuracy. But when the player does not have the same move time pattern, then there are dozen of blunders and mistakes on average.

This is all very subjective, it depends on the game/games in which you are claiming they cheated,

I agree with you, the rating graph does look suspicious - although that is far away from proof.

In the future instead of the investigation you did I can only recommend to just report them, let Irwin do his job and simply move on with your life. It is better for your peace of mind.

If I could see the games I could tell whether he is an engine cheater or not. Does the account have a bullet rating? Is there a
much higher classical or rapid rating?

Do they play 3+0 games sometimes? It is not impossible to cheat at 3+0 but it is quite uncomfortable so most cheaters prefer longer time controls.

I believe that a big difference between blitz and classical rating with the classical rating much higher is a clear sign of cheating.
Accounts without a blitz or bullet rating but a high classical rating are always very suspicious. Nearly every active confirmed titled player on this site has some blitz rating.

Ask Irwin:

github.com/clarkerubber/irwin

@mrbasso said in #7:
> If I could see the games I could tell whether he is an engine cheater or not. Does the account have a bullet rating? Is there a
> much higher classical or rapid rating?
>
> Do they play 3+0 games sometimes? It is not impossible to cheat at 3+0 but it is quite uncomfortable so most cheaters prefer longer time controls.
>
> I believe that a big difference between blitz and classical rating with the classical rating much higher is a clear sign of cheating.
> Accounts without a blitz or bullet rating but a high classical rating are always very suspicious. Nearly every active confirmed titled player on this site has some blitz rating.

All of the games are 5+0 blitz. I'll have a look at Irwin, but I think for the piece of my mind I'll just not think about it. In case that person is cheating, he will waste his time while I learn about the game. In case he does not cheat, very impressive and good for him/her!

Update:





I guess my hunch was (seemingly) right after all :D!