TT.2022: strange probabilities

Please stick to the my point. It's fine to apologise. But, you aren't answering the question. Why did you spread that misinformation? The only logical reason is that you were accusing him, and using that as evidence. So, admit you're accusing people.

For those who have difficulty hearing ( or reading ), I repeat.
I don't blame anyone for anything ( !!! )
.
.
Please don't clutter this topic with spam...

Except, you did. You clearly speak misinformation about snowlord in order to increase suspicion that he is a cheater. So, you're just in bad faith when you say you're not 'blaming' (the word is 'accusing' though).
You are accusing and you need to deal with the fact that these people are certainly going to be innocent at least some of the time. Therefore, you are certainly falsely accusing at least some human beings who don't deserve it. Do you really have nothing to say about this at all, except dumb denial?

Don't test the strength of my nervous system : )
For you personally: complete ignore.
I'm stopping replying to your messages here.

Then, you're just absolutely refusing to consider what you're really doing, and what the possible consequences of it are. That's a shame.

That account is from chess.com Kramnik...

Is this the profile you're talking about Kramnik?
https://lichess.org/@/gogotaniyeIt's still open.
Then there's this one, which is closed. But I'm not seeing lyrics, photos, links etc. Are you sure you're not just a little bit confused?
https://lichess.org/@/snowlord

@MichaelKagansky OK, I see, your WDL model does not take into account 2 things:

1. Time control. Rating can be collected at one control and the game you estimate result for in other time control. And you ignore it. E.g. Kramnik would lose in 3+0 to almost anybody, while would put up a fight in 3+2.

2. Player strength. Only current ELO. Something like average rating at least of some sliding window would be way better. Because rating can move +-100 pts during months if user plays on website regularly, even for players like Magnus, who is considered to be very stable player.

@VladimirKramnik

> I have checked this account on lichess, there is NOT A SLIGHTEST INDICATION the account is someone else's than yours. There is your name,photo, link to your fide personal page,even your lyrics there. Lichess doesn't open accounts with someone else's data without checking

1. What? Anyone can create account on lichess with your avatar, name, photo and link to your fide profile at any moment. Do you really think that lichess manually checks all unverified accounts users create? Do you really think that? And even if they were, what should they do for some unverified account without any third-partly reports or complains? Ask all users their passport scans during registration?

2. Also you provided the link to the chess.com, not lichess :facepalm:

P.S. Mr. Kramnik, maybe you do not know well enough how internet "works", is it the case? It would explain your confusion and logical mistakes, e.g. you wouldn't read gazeta_ru, lenta_ru and other "RussiaToday" "news" sites (they were visible on YOUR public stream in your main browser bookmarks), but instead you would have chosen more reliable news websites with better reputation on the internet.

> Concerning the other one snowlord, no prove now whose account was it, but since exactly the same nick, logical to assume it was his account until it is proven the opposite.

Yes, with such a UNIQUE nickname like SnowLord he probably got himself [email protected], snowlord.com and all other snowlords, it's hard to imagine somebody would choose snow + lord, too. Hmm, wait a bit, maybe the combination of this 2 words is a liiiittle bit more popular:
https://duckduckgo.com/?t=ffab&q=snowlord

@VladimirKramnik
You just went on a completely bizarre attack because you were confused. Then, you decided not to apologise. This is why you're having trouble getting any respect. You aren't taking accountability for your actions, you aren't accepting when you're wrong.

@bifree
Chess^com ( or FIDE ) does not separately calculate ratings for different 
controls. My WDL-model uses only Chess^com ( or FIDE ) rating at the input.
.
.
For a weak ( by rating ) player (3+1)-control almost always helps 
against real blitz monsters.
.
"Nuances" can work in any direction. For example:
1 / 1.000.000 >>> 1 / 500.000 ---- OR ---- 1 / 1.000.000 >>> 1 / 2.000.000
.
.
Don't divert the public's attention to unsuitable objects.
( ? ) Explain what it was ( in just one 2022 year ):
.
0 / 5000000
1 / 1250000 ( 1 out of 1.250.000 TT-tournaments / 1 per 12.500 years )
1 / 1000000
1 / 312500
1 / 71428
1 / 69444
1 / 56818
1 / 31645
1 / 22222
1 / 18450
1 / 16949
1 / 13812

@bifree

Important note.
.
.
Chess^com monitors the TT tournament as follows:

20 players significantly increased their rating ( +50 ... +100 )

20 players significantly lowered their rating ( -50 ... -100 )

OK... 98% that this was supposed to happen : )
.
.
The WDL-model allows us to determine the probability that accompanies 
such progress in the ranking. Here 1/100 or 1/100.000 are two big differences.

Are you talking to me or someone else

Are y'all good because it seems like y'all just like arguing to my people to apologize but I don't got nothing to do with this so just leave me alone I'm just trying to play chess and like trying to start with other people on here Okay

Yo are y'all going to battle or what we should really play chess because like chess is the best game you can play Also when I go home I can play so but just don't call me don't call me at no 3:00 be like bro are you ready because I mean I'm just going to ignore all of your calls here I will give you the phone number 856-308-1921

@MichaelKagansky

> Chess^com ( or FIDE ) does not separately calculate ratings for different
controls. My WDL-model uses only Chess^com ( or FIDE ) rating at the input.

Don't you consider it to be a problem? 3+0 and 5+3 are completely different games, and your method does not take it into account (nor chess.com rating does). I would play these 2 time controls with drastically different results, Kramnik, I am sure, also would.

Idea for improvement: use rating only in relatively similar time control. And make it based on some period of time.

Nakamura and other strong GMs showed that +200 rating is easily achievable if strong player wants to actually gain rating by farming. Another example: Dubov does not care for rating or even TT games results, but if he had this task, he would gain +200 in a short time. So, that's why guys like Chess Ray does not believe in chess.com rating that much, it changes and does not reflect the strength that good.

> Don't divert the public's attention to unsuitable objects.
> ( ? ) Explain what it was ( in just one 2022 year ):

You want me to explain? Why? If you accuse these people, make a report and let chess.com think about it. If you don't accuse, what explanation you want? All we know, the any results are mathematically possible and for one random case I randomly took from your images, I showed that it is not only possible, but expected. Now we know about your model (just use chess.com rating and estimate outcome based on it in possibly different time control), and it also makes the case less substantial in the first place.

If you assume that your "suspect" players are 200 points lower rated than the opponents then that is approximately a 25% chance to win each game (well, to take one out of four so two losses and two draws probably form a big part of that 25%). In some games they will have a better chance because they are closer than 200 points and in some they will have a worse chance, but let's go with a 200 point average. In that case then the chance of an individual player getting 9+/11 is 0.0126% (at 1/7936 it is bit higher than your number of 00117%). That assumes every game is decisive, on the assumption that two draws have roughly the same possibility as a win or a loss, so it added up the likelihood of 9/11 (with 55 possible distributions of the two losses), 10/11 (with 11 possible distributions of the loss) and 11/11 (only one possible distribution). It did not include the likelihoods of 9.5/11 and 10.5/11 and it did not take into account the possible distributions of draws, so 0.0126% may be a bit low.

In the norm events I've worked the norm winners generally did very well against those rated close to them or below and still reasonably well against those rated higher. Not every game will be against super GMs so instead of playing 11 very very strong opponents it will be a bit less, but let's still assume 11 very very strong opponents.

If 400 such players were in an event then that is a 4.92% chance of at least one of them going 9+/11 (better than the 4.57% chance using 0.0117%) for a 91.96% chance that 50 such tournaments would have at least one 9+/11.

How many players (per TT) below (but still close to) 2750 are you considering? How many TT events are there in a year?

For that matter chess.com has 16 other accounts starting with snowlord and two more with snowlord as their descriptive name.

Ill lay later at 12 17