The New "Evidence" Against Hans Niemann is Average Centipawn Loss analysis

A data analyst and chess enthusiast Rafael Milk has been posting very confidently that he has found irrefutable evidence that Hans Niemann was not playing clean.

When you look at his analysis it basically all relies on analyses the linear relation between average centipawn loss and Elo. He assumes it to be true and then proceeds to show how Niemann's results disprove his assumption.

He doesn't show us an rigorous empirical evidence that his average centipawn loss theory is true or neither does he show us any tolerance levels or limits for the linear regression he is doing.

He also doesn't use the timescale. He assumes the timeline has no impact on the analysis. Hans Niemann for instance spent over a year at 2400+

It seems that it is an interesting approach but may be confirmation bias, since he creates a rule of thumb and says Niemann breaks the rule without showing that the rule works in on legit cases and doesn't have legit exceptions.

Chess drama continues...

www.youtube.com/watch?v=Q5nEFaRdwZY&t=1s

Rafael Milk's analysis is "sour".

Many of the smartest data analysts will have opinions. Many of the greatest chess players... And so on... And in the end, none of these opinions will mean anything.

These people decide:
"
The members are WIM Salomeja Zaksaite of Lithuania (chairperson), a scholar of criminal law and a criminologist; Vinzent Geeraets, an assistant professor in the department of Legal Theory and History at Vrije Universiteit Amsterdam; and Klaus Deventer, an international arbiter and substitute member of the FIDE Arbiters Commission.
"

My opinion: sorry, but that panel is completely under qualified. I am very disappointed with FIDE again. This is not a panel that screams "we are putting all our energy into this". Maybe it's time for chess to move on from FIDE...

I reviewed all three of Milk's videos on the Big Drama and his methods seem sound.

Whether or not his actual methodology was correct remains to be seen as many others now endeavor to replicate and expand upon his findings ...

Using two of the most elementary statistical metrics, the average and the variance, he first discovers the 'structure' of these two measures across the entire space of around 8000 games played by many top GM's ...

This structure reveals the manner or pattern in which the 'average' GM in the study improves over time.

All GM's in the study became more and more accurate in terms of CPL and also more and more 'consistent' in their moves as they journeyed from 2300 strength to 2700 ELO ...

All of the GM's in the study except a single outlier ... and guess who that was ...

even if the moves match computer moves, there needs to be a second element: how does the player get information on the computer suggestion? until I hear a plausible answer, I cannot conclude someone is cheating

@sparowe14 said in #5:
> even if the moves match computer moves, there needs to be a second element: how does the player get information on the computer suggestion? until I hear a plausible answer, I cannot conclude someone is cheating

It seems that this angle of attack is not required. The attempt to use "evidence" from online matches a few years ago in order to destroy his reputation now appears to be the chosen path.

The fact that 0% of his FIDE ELO rating comes from chess.com seems irrelevant. As does the fact that he gained those points largely via OTB tournament play over a number of years. The possibility that he has cheated all the way to his current rating is fantastical.

@MidiChlorianCount said in #6:

> The possibility that he has cheated all the way to his current rating is fantastical.

This is of course correct, but unfortunately you'll never convince those who taken a position against Niemann, come what may, even with reasoned argument. *Especially* with reasoned argument, perhaps. Some people love blood sports.

@MidiChlorianCount said in #6:
> It seems that this angle of attack is not required. The attempt to use "evidence" from online matches a few years ago in order to destroy his reputation now appears to be the chosen path.
>
> The fact that 0% of his FIDE ELO rating comes from chess.com seems irrelevant. As does the fact that he gained those points largely via OTB tournament play over a number of years. The possibility that he has cheated all the way to his current rating is fantastical.

You didn’t watch the video? The conclusion is that Niemanns accurate rating is around 2500. There’s no statistical anomalies in his OTB play up till the year 2020 when his Elo approached 2500. The rise since then is statistically anomolous.

@ClayAndSilence what would be your perceivedly "reasoned" arguments?

And what are the reasons some people hate Magnus so much?

@gravling said in #8:
> You didn’t watch the video? The conclusion is that Niemanns accurate rating is around 2500. There’s no statistical anomalies in his OTB play up till the year 2020 when his Elo approached 2500. The rise since then is statistically anomolous.

So yes, according to chess.com he stopped cheating online in 2020. And according to this video - at the same time - he takes up cheating OTB. Interesting stuff.