1. In suspect games, find out what percentage of Niemann's moves are the same as the moves that, e.g., Stockfish would make in the same position. Call this the "coincidence rate."

2. Compare this with the coincidence rates that resulted when similarly ranked players who are not suspected of cheating played against the same opponents.

Do the same thing with other chess engines that Niemann may have used. (If the various engines almost always come up with the same moves, this is unnecessary.)

3. If the coincidence rates in Niemann's games are significantly higher than those of the players he is being compared with, that would be prima facie evidence that he is getting help from an engine.

A *statistically significant* comparison, i.e., one showing that Niemann's coincidence rate is higher than other players' by more than the average variation among non-cheating players, which can be ascribed to chance, would require checking a lot of games.

They said that Hans had the <fastest ever> increase in OTB strength between age 11-19 (faster than Fischer and Carlsen..), and they showed that he had <plateaus of 4.5 years> between ages 13-17+ which is <extremely unlikely> for someone who is supposed to be one of the most talented players ever..

While these two conclusions raise great suspicions regarding Hans cheating OTB, they are not strong enough by themselves (as chesscom themselves said).

--

The rest of their OTB analysis showed nothing special, but my problem with that is:

1. Those analysis were based on their Strength-evaluation, and they said that that algorithm is <optimized for online blitz games> and <not for OTB> games.

2. They dismissed significant work published others on the net, claiming their methods are below their standards, even though they <never checked their own algorithms for OTB games>, and the people they dismissed <did check> their own algorithm with OTB data!

3. They ignored published information about Hans performing way better on broadcast events, that could add an important dimension to their anlysis.

(1)
Very serious and interesting work by <Rafael Milk> from Brazil, data analyst and a strong chess amateur (and streamer):

He used av. centipawns to show that during his rise from 2300 to 2700, Niemann's chess strength (almost) did not rise, and if that is true, then his level just <jumped> from 2300 to 2700 in <one day>, and his rating just needed to catch-up with that level.

He then showed that his av. centipawn strength measurement method, is inline with players rating: so there is a <very good correlation> between player rating and his centipawn strength. That is an extremely <critical> information for his analysis, and it something the <chesscom did not do> with their (online blitz based) strength analysis (they did not show strength-rating correlation).

Here he shows the correlation, and also presents that this correlation is kept during the progress of known top players (other than Niemann of course...):

https://youtu.be/Q5nEFaRdwZY?t=147

Based on the stength-rating correlation that he found, he concluded that during Nieman's rise from 2300 to 2700, his av. strength <remained around 2500 level>. The most reasonable explanation for that, is that Hans used engine support only on critical moments, while most of the moves he played by himself (with his true strength close to 2500 rating).

Here are his 2 full videos:

(2)
Work by <FM Yosha Iglesias>: The analysis were based on chessbase game analysis feature, that gives game strength-score compared with the engine. chesscom insinuated that this method is below their standard, but again, they <never tested> their strength-algo with OTB games (and maybe chessbase did test theirs..), so I don't see reason to dismiss Yosha's analysis.

And based on those strength analysis, Niemann had <extreme anomalies>, in number of perfect games (20 times more than everybody else..), and his 5 best tourneys were better than everybody else including Fischer.

She explained that cheaters who knew about Kenneth Regan's method (that is used by fide), could easily <bypass it> if they played different games at different levels. Plus his analysis ignored the time per move (surprisingly naive algorithm..):

https://youtu.be/hFcSwSIw9nE?t=941

She also said that the chessbase-analysis is <not a known method>, so cheaters could not take measures to bypass it. Interesting observation..

She also showed statistics where Niemann has the <highest> number of top-3 engine moves compared with other top players, but the <lowest> av. centipawn score, and that is another indication that he picked engine moves on specific moments, and the rest of the game he played on his own (which is enough if u get a winning position..):

https://youtu.be/hFcSwSIw9nE?t=187

Here are her 2 full videos:

https://www.youtube.com/watch?v=jfP...
https://www.youtube.com/watch?v=hFc...

---

(3)

And last bot not least, is data showing that between 3/2019 - 11/2020, Niemann played at 2400 level when the tourney was not broadcast, and at <2610> level when it was broadcast:

https://twitter.com/ATL_kings/statu...

and that there was no significant difference in strength between the two cases:

https://twitter.com/ATL_kings/statu...

From that same day's broadcast I remember this from Wesley So talking after his five hour draw with Aronian (So vs Aronian, 2022) :

https://www.youtube.com/watch?v=TkU...

A *statistically significant* comparison, i.e., one showing that Niemann's coincidence rate is higher than other players' by more than the average variation among non-cheating players, which can be ascribed to chance, would require checking a lot of games.>

On the other hand, Hans has admitted to cheating in the past, but it seems that he hasn't been completely honest about it yet. I see the point in «information, that contradicts his statement». This goes to whether I think Chess.com is doing the right thing.

It seems, that Chess.com has always been especially flexible/lenient when it has to do with titled players. This is something to juxtapose to 'There should be no leeway for cheaters.' Also, juxtapose 'Magnus throws a fit'. I can't endorse a witch hunt. Aren't everyone's games auto reviewed by their anti-cheat system? ? Was it sleeping all this time and only after Magnus throws a tantrum that they went back & retroactively scanned Hans' past games to dig up more shade on him?

Either way chess dot com don't come off this looking as squeaky clean as they might want.

[So] misses a simple mate???>

Sure, they're already exhausted from a grueling game, and don't even have 10 seconds of quiet time to think about the position.

I remember one time So was interviewed to analyze a game and immediately asked that an engine be turned on.

Certainly didn't mean he's a cheater needing help, but rather that he wants to make sure he doesn't suggest a non-workable move to the audience.

I cheated once when I was 12 and 16 but in unrated games.

I cheated at least 100 times when I was 12, 14, 16, 17 in rated games but never for money.

I cheated at least 100 times when I was 12, 14, 16 and 17 in rated games and for money but never when I streamed.

I cheated at least 100 times when I was 12, 14, 16 and 17 in rated games and for money while streaming at least 25 times but never against GMs.

I cheated at least 100 times when I was 12, 14, 16 and 17 in rated games and for money while streaming at least 25 times and against GMs but never OTB.

I swear.

But we always knew he cheated over more than one game when he was 16 - given his first public confession.

In fact, though people credit Carlsen with bringing the cheating scandal to the fore - but it's really Niemann who did so with his initial confession. He pulled it out of the shadowy world of innuendo and put it squarely on the table.

Without that confessional interview we might still be debating exactly why Carlsen pulled out of Sinquefield.

< Sep 6 > Hans declares he cheated online in 2020

Nonsense. Have you done the analysis?

<20, 25% of GMs and IMs have cheated online, that's my guess. . . . But however, if you ask me what percentage of these people cheated in real life, it's much lower. It's less than 1%.>

<GM Ben Finegold on the Hans Niemann WSJ Article>

https://www.youtube.com/watch?v=df6...

Not quite as repetitious this time.

Basically it seems Ben and I could just tag team - his views are quite similar to mine, though I might be a little easier on Niemann since I'm not quite as surprised that the online cheating stopped in 2020.

One other thing, a possible difference in our viewpoints - Ben says "<chess.com> doesn't do OTB", yet they did have this statement in their report:

<Despite these potential suspicions, as shown below in Section VIII, an in-depth review of Hans’ OTB games using Chess.com’s statistical methods revealed aggregate patterns of play that, while interesting, are possible for a rising player approaching 2700.>