Cheating in chess

[NickFaulks]

Also for that matter, there are players of lower ratings who can choose the same moves as titled players perhaps 18 times out of 20. The 19th and 20th moves are invariably fatal of course. But it's a completely plausible explanation as to why players can go 300 points above their published rating for short periods, such as the length of a tournament.

It is claimed that Prof Regan's system deals with this. However, all that Yuri can reveal about it is that it is far too sophisticated for our small brains to understand.

[LawrenceCooper]

It may be this thread from 5th June from Emil:

Thanks, I'm sure you have found it, and presume there is more.

Hiding behind some illogical provisions (you can't suspect anyone based only on the strength of their play. Why???)

What Sutovsky has never managed to grasp is that there is a distinction between suspicion and public denunciation.

Agreed. His consistent over-reaction every time he hears of a cheating accusation can't be helpful for those involved not to mention his disdain for lower rated and older players being capable of performing above their ratings

[Brian Towers]

But it's a completely plausible explanation as to why players can go 300 points above their published rating for short periods, such as the length of a tournament.

And, of course, things like bereavement, illness, divorce, job loss can cause similar movements in the opposite direction which can last a lot longer than one tournament. If you play against such a player you are likely to have a result well above your playing strength never mind rating.

Absentmindedly I've overlooked age. Speaking from experience, an older player can play a lot worse in round 5 of a weekender than round 1. As you get older taking steps to overcome this is probably more important than opening preparation.

[Roger de Coverly]

Specifically, outperforming one's rating by 300 points or more over the course of a tournament of nine rounds or more?

I don't think it's that difficult or unusual. I scored 50% at the last London Classic FIDE Open with a rating performance according to chess-results of 2162 against a starting rating of 1921, so that's +241. I certainly missed a win in round 8 and possibly a draw in round 2, which would have boosted my performance towards + 300.

The following month I won the Major in the 4NCL Harrogate with 4.5/5. That was a +290 performance of 2259 against 169.

At Basingstoke last month, I also scored 50%. That was for a performance of 1753 against a start rating of 1943, so -190.

I'd imagine that improving Juniors frequently get +300 results or better.

[JustinHorton]

Right, but:

(a) that's no actual examples

(b) juniors are another matter entirely

(c) I actually wondered how rare or otherwise it was, because I'm trying to see if Sutovsky is reasonable or otherwise in thinking it's proper cause for suspicion. So I wondered if anybody had actually produced any stats.

[JustinHorton]

What I'm asking, in part, is this: is there any actual statistical yardstick for considering a given level of over-performance to be intrinsically worthy of suspicion?

[Mick Norris]

There's no doubt something somewhere from Ken Regan

His 2013 article has the following

Now suppose we have a tournament of 100 players, and we select some to test for cheating. We might select the first prize winner, or the top three or all those in a tie for first, but chances are they were among the highest rated players to begin with. Instead, as we glance over the tournament crosstable we may notice a remarkably high Plus Figure, like +250. This means the player had enough wins and draws to meet the expectation of a player rated 250 points higher. The player may still have finished in the bottom half of the tournament with more losses than wins, but if the player was among the lowest rated that could still be a highly “plus” performance and even win a so-called rating class prize. Some players are even alleged to lose on purpose in one tournament so that their rating will be below the cutoff for a class prize in the next, a practice called “sandbagging.”

So we test our Plus Player, and we get a z-score of 2.50, meaning odds of 160-to-1 against it happening “by chance,” and certainly a significant statistical outcome by civil convention. Then what do we do? The answer is: absolutely nothing. Chances are that player was just the shiny marble.

[Roger de Coverly]

is there any actual statistical yardstick for considering a given level of over-performance to be intrinsically worthy of suspicion?

Sutovsky is probably more concerned about protecting GM exclusiveness by casting doubt on 2300 FMs playing to a 2600 standard, than he would be about a 1700 player making 2000 standard performances. For a filter at a lower level, look at the winners of (W-We) rating prizes as they are the ones with the greatest out performance.

The mathematics underlying the Elo system assumes variability of performance with k factors expressing the certainty or otherwise that a good performance is indicative of an improved standard rather than just a fluctuation.

(edit) As pointed out in the article above, a +250 result doesn't happen by chance. The Elo hypothesis is that you now have information to suggest that your previous estimate of the player's strength, the Elo rating, is incorrect and should be adjusted upwards. (/edit)

[IM Jack Rudd]

I think I have an example: http://chess-results.com/tnr257154.aspx ... g=30&snr=7

[Roger de Coverly]

I think I have an example

A 2508 performance should be well within the range of expected outcomes for a player with an IM title. If automatic screening ever became established, false positives would be eliminated by checking against a player's peak rather than current rating.

[Brian Towers]

There is a problem with statistical analysis which most non-statisticians don't understand.

The standard test "95% confidence limit" basically looks at some data and asks "is the probability of getting this result by chance less than 1 in 20". If the answer is "yes" then the result is flagged as statistically significant.

The obvious problem with this approach is that "1 in 20" events happen about once every 20 times. If you have 100 players playing in a chess tournament then on average about 5 of them should get a "1 in 20" performance.

Changing your level of statistical significance to, say, 1 in 100 doesn't help. 1 in 100 events occur on average about once every 100 tests.

The old saw "correlation does not equal causation" applies. The statistics can tell you where it is worth looking but you need hard evidence to draw meaningful conclusions. Suspicious behaviour is a much better indicator than outperformance.

[IM Jack Rudd]

Good old conditional probability. We can, with the help of the FIDE rating system's assumptions, assign some reasonable probability to "if a player is not cheating, what is the probability of their getting a TPR of 300 or more above their rating?", but what we actually want to know is "if a player is getting a TPR of 300 or more above their rating, what is the probability that they are not cheating?".

[Brian Towers]

Q: What is the probability that you and I have the same birthday?
A: 1 in 365 (ignoring leap years)

Q: In a room of 23 people what is the probability that there are at least 2 people who share the same birthday?
A: Close to 1 in 2

Q: In a room of 100 people what is the probability that there are at least 2 people who share the same birthday?
A: Very close to 1

[Roger de Coverly]

but what we actually want to know is "if a player is getting a TPR of 300 or more above their rating, what is the probability that they are not cheating?".

In an over the board tournament, they are under almost continuous observation by the opponent, the other players and sometimes even the arbiters. Any move matching analysis should take this into account.

That's what the ACC have been doing, is it not? There has been some other suspicion such as a phone in the loo, bluetooth headphones etc.

https://www.chess.com/news/view/life-ti ... r-cheating

[Kevin Thurlow]

"Agreed. His consistent over-reaction every time he hears of a cheating accusation can't be helpful for those involved not to mention his disdain for lower rated and older players being capable of performing above their ratings"

Doubly agreed. I now worry if I get a good position against a strong player (obviously a rare event) that I'll get accused of cheating!