TT.2022: strange probabilities

Here are 20 results with a ( < 1/1500 )-chance

Are such “anomalies” a signal for any Chess^com reaction ( ? )
A short comment from the TT-organizers would be interesting...

Hello Michael!

Did you calculate the probability of such performances happening all in the same year? would be interesting to know

with my best regards,

Marta

It is impossible to count EVERYTHING...

But... There are 104-106 TT-tournaments a year...

And if someone has overcome ( 1 / 100000 ) - he has already

completed your “norm for sensations” for the next 1000 years : )

Vladimir, why are you using a fake account to make accusations?

For fans of fakes:

https://ratings.fide.com/profile/14100967

https://www.iccf.com/player?id=941222&tab=3

https://lichess.org/@/MikeKagansky

I see that the topic does not interest you...
Please don't litter it with trash...

It wouldn't be the first time Vladimir used an account of a titled player, would it? That's what he was caught cheating for.

1. How do you calculate the probability, what is the formula you use?
2. Is it based only on ELO of players on the website?
3. Can you:
   1. formulate the hypothesis,
   2. share the model,
   3. choose criteria for it to be true or false, and
   4. share the calculation of actual probability that your hypothesis it true?

Without it, all these numbers state nothing.

The same goes for Kramnik with his "statistics", when he actually only "calculates" average of several values of entertaining "accuracy" number, which he does not even know about how it's calculated.

I can understand the importance of "nuances" in methodology 
if we are dealing with probabilities like 1/5, 1/20, 1/100...

In the case of 1/2000 or 1/20000, using “nuances” 
you can get 1/1500 or 1/15000...

This is not the right place for a detailed description

my software works... Look also here:

https://twitter.com/KaganskyMike

Your twitter does not show anything of this.

And I did not got an answer. Your "software works" but you cannot share any information about real science you used? Like what is your hypothesis? What level of the game outcome probability should show somebody is a cheater according to your model? Does this level exist? What is your model? What criteria should your data meet to draw conclusions about something? How can you check your hypothesis is true of false within some probability range?

Maybe, you are a bit surprised why almost nobody supports you in this, and maybe you think that your data means something important, but it is not, until you can share any actual math and statistics approaches you used (if any). That is how science work, read about falsifiability, that is what all your pictures in twitter lack.

I don't blame anyone for anything.

.
I show "anomalies" and indicative calculations.

All input data is taken from the Chess^com and FIDE^com websites:

ratings, dates, tournament results, game results, etc.
.
So far no one has refuted my “calculations”.

No one showed a competitive method.

This means that there are no big claims about the probabilities.

Nobody said that 1/2000 is nonsense and in fact there is 1/20, for example.
.
I'm not trying to "bring science to the people".
I'm wondering what the Chess^com-reaction is in cases 
of players "overcoming" such low probabilities:

.
( 1 ) None

( 2 ) Further analysis of the player's Chess^com-results

( 3 ) Sanctions

The documentation will be in several parts.

I hope that the number of "questions about methodology" will decrease...

> So far no one has refuted my “calculations”.

Why do you think, someone should refute it? It does not state something, like this player was using third-party assistance with such probability according to this model and these criteria, I do not see it.

It's hard to "refute" raw numbers made by addition and division. Refuting the drawn conclusions would be possible if the math would be provided.
What is the final conclusion of your calculations?

> No one showed a competitive method.

Method to do what exactly?

1. Let's take 1 Titled Tuesday.
2. Let's imagine all games are played COMPLETELY fair, without third-party assistance and without other people's accounts.

Can you estimate the probability that one or more participant from this single TT, under the provided assumptions, would show the result better than 1/1500 chance, according to your way of calculation?

Let's take some example of yours and check. I will assume you took all the ratings and other stuff correctly.
I took randomly: "Mark Plotkin" image.
It's a real person but we can also consider him as just general strong IM in many cases.
1. He has FIDE-blitz rating 2401. According to fide.com his OTB blitz rating is still growing EACH offline OTB event for the last 10 years. Does this rating reflect his level? I do not know, what your model says, it does? Let's imaging it does.
2. His chess.com rating is 2752. Do you consider it to be valid or not? 350 points gap. Is it fine for such high level, Hikaru, Magnus and others also have much higher chess.com rating than OTB. In your model, is it more probably that it's a correct rating or not? I think it's a fine gap, easy to imagine that it's real.
3. You show that he once got 9 points. And the probability of getting 9 or more points according to your calculation is 0.0117%.
Sound incredible, right? Let's check.
4. There are more than 100 TT each year. Such strong IM was almost certainly playing for several years on average, including pandemic, so hundreds of chances to take. Let's imagine he was playing only once in 4 tournaments and took 100 TT in total (we can check it for Mr. Plotkin but I speak about general case of strong IM).
5. So, 100 tournaments, now the probability to over-perform in at least one tournament and get 9+ points is almost 1%, and it's only for him!
6. In each TT we have ~500 players. About 100 of them can be at this level of strong IM or weak GM. EACH tournament.
7. Talking into account that according to my previous assumption the participation ratio is 0.25, it means there are like at least 400 players who can be considered in the same shoes as Mr. Plotkin (rough estimation, of course, can be checked and corrected).
8. The probability that some of 400 people of this play level would over-perform in their TT and take 9+ points is more than 98% now!

Conclusion:
Almost certainly we will see such cases as Mr. Plotkin (with 98%+ probability). Not getting such cases would be highly unlikely. Furthermore, almost certainly we would have a bunch such cases at his level. Even assuming everyone is everything is fair and their rating is precisely reflecting they level (both these assumptions are not true). Also assuming your rating estimations and probability calculations are right.
Can we say "these Plotkins" should be punished for something? I do not think so, do you?

So, what is your point, then?