So if you're comparing move accuracy to rating, how do you tell the difference between a cheater and an underrated player? Or the difference between a consistent cheater and an appropriately rated player. This analysis seems to be asking the question "is a player making moves consistent with their current rating?" In the long run, the answer to this should be "yes" by definition, regardless of whether the player is cheating or not.
Indeed - it also completely ignores the fact that a competent cheater would make an active effort to smooth his results when graphed. It would be insane to make the best move, all the time, in every game, and come in #1. A person's goal may be just to slowly raise his rating to get into the higher tournaments and then to take a good payout and move on with life.
I think there are things that come along with an underrated player. Let's look at niemann for example, since his chesscom account is open, which I assume is after his previous ban, he played more than 2500 bullet games and 1000+ blitz games. Für bullet it's already really hard to cheat, but overall you would expect chesscom to be able to spot and than ban him again if he cheated in these games. Everyone is talking about all these red flags, but there are so many green flags too clearly indicating that Niemann has always been an extremely good player. Did he cheat? I don't know, but certainly right now there isn't that much supporting the theory other than Carlsens gut feeling. Caruana even looked at some of the games with very high accuracy and in most of them found all moves Niemann played relatively natural, another GM I saw analysed 1 of the games where apparently he played like 30 top 1 engine moves in a row and came to the conclusions that a high rated gm would probably found all or close to all of these moves even in a blitz game and Niemann had much more time to find them in a classic game. the problem is that many people have a biased approach, instead of checking if Hans is cheating or not, they approach it trying to proof that he cheated or trying to disproof it, instead of collecting all the facts and then maybe drawing a conclusion or maybe understanding that the evidence isn't conclusive.
@@LunnarisLP I understand your point, and I don't think there is any possible way that I could use any device and beat Magnus or Hikaru. It can be true that Hans is an extremely powerful player and also that he is cheating. In fact, I would expect that to be the most likely case. So his other stats, like bullet chess, would show that. But the number of times he has had 100% or 90%+ engine top move correlation is too insane to be ignored. I do not understand those data in any way other than he cheated in many (but not all) of those games. A 45 move game against one of the top GMs in the world with a 100% engine correlation? That MUST be cheating.
@@LunnarisLP My view is that we should assume that if a player wants to cheat, they will, and that we will never be able to prove it. Regan's algorithm has been put forward as the leading statistical method for cheat detection, but as described here, it can't detect cheating even in theory. Niemann is a cheater by his own admission. He has displayed a pattern of cheating over the course of 5 years including in games in paid events, most recently in 2020. We will never know whether he stopped cheating in 2020 or just got better at it. People can decide for themselves whether they want to play against him or invite him to tournaments, but it's a completely reasonable decision to never want to play against him again.
well said.. exactly what I was thinking as I watched.. the analysis seems to be missing a key point if I understood it correctly.. It seems to assume that this guys aggregate rating and most of his matches come from playing honestly, and thus it's analyzing recent matches to see if there's outlier numbers/ROI.. but if he's been cheating his whole career.. (which there's seems a lot of belief in the chess world, as well as his admissions that he did this when he was 16 (only three years ago), then this entire analysis seems null and void
There's a 'not seeing the forest for the trees' element to many folks discussion of Regan's model. His analysis is extremely rigorous and well-executed, but it is limited in scope: It can detect cheating very well as long the cheating meets certain expectations. This is inevitable considering that his statistical model requires a very large data set and, at it's core, compares deviation of sample from expected performance with respect to a leading engine's main line. Plainly, it works as long as the cheater cheats in a predictable way over a certain period of time. A wise cheater with an understanding of such analysis could likely go undetected by strategically limiting his "indiscretions." Overall, Regan developed an excellent tool for *helping* to keep Chess fair, but it cannot do on it's own...especially at the highest levels. Unfortunately the other tools in anti-cheating "toolbox" are clearly still in early stages of development.
Precisely. It's a broad level analysis of the full data set basically to detect the equivalent of a statistical rising tide caused by systematic cheating. It won't detect cheating if it is dictated by other contextual or external conditions. For example, if the method is restricted and doesn't have equal opportunity to be employed then you could have a data set where of all games the opportunity to cheat wasn't large enough to create a significant statistical deviation across the whole set.
Considering that Hans has 10 games on chessbase with 100% score and a bunch more over 90% it is puzzling to me that Regan's algorithm didn't catch anything wrong. No other player even has one 100% score and very few have one over 90. This is super fishy.
@@tristan6066 Ok, I haven't investigated anything I just watched the Yosha channel video. She said the highest correlation score before Hans was 98% and that player got busted for cheating. It seems really odd that Hans would play that many perfect games even if he was cheating. He is basically narc'ing on himself doing that.
@@gametime2473 She compared his individual games with tournaments or sometimes even multiple years for other players which makes no sense. And ignored the main reason Hans was able to play this high accuracy games, it was because he was playing way lower rated players, which he did because his own rating back then was much lower than his actual strength. The fact he already was that strong at the end of 2020 when his fide classic rating started to go up is clearly visible by e.g. his chesscom rating already peaking there for bullet and blitz, in case of bullet chess he was ranked top 5 at some point. Given the amount of games played I find it hard that he was able to keep the rating while cheating and on top of that not getting detected especially given chesscom likely paid extra attention to a guy they had just banned like a year before or sth.
"You wouldn't cheat just one game" Well, for quite a lot of people, beating the world champ would really be up there with "You would only cheat to win the tournament or financial gain". Other examples where your conclusion fall flat is if you basically the 2nd best player of the tournament and you only need the cheating against that one player above you. Last but not least some people has a nemesis, pure emotional hate/dislike of another person.
@@Tarotainment he never said Hans was 2nd best. He said if you cheat against the 2nd best player, you'd only need to cheat once, especially if you can beat everyone else. If you're the third best player for example.
Nothing indiciates Hans cheated against Magnus. It was not Nieman playing super great there. Magnus had an Engine correlation of 41% in this game, which for him is just super bad.
The thing is that if you are already a good player, the crucial times you need to cheat is very low. Statistical analysis will only see cheating patterns if the cheater in question cheats a lot. A cheater cheating one or two moves in a critical position in one or two important games in a tournament won't be noticed in such analysis.
@@kartikeyatiwari2502 No. By that logic, statistical analysis of checking for cheating is a useless tool. If you want to find a cheater, you'll need to find actual evidence. (ie, find tech and other signalling solutions.) And in absence of this, suspecting anyone is a lost cause.
@@Asrudin "By that logic, statistical analysis of checking for cheating is a useless tool. If you want to find a cheater, you'll need to find actual evidence. " You actually need to find evidence the case in the video where he suspected someone of cheating. They didn't say "Ken Regan says statistically you are guilty.... so..." They watched him carefully after his statistics were shown to be suspicious. Useless no, smoking gun evidence fuuuck no. Like a Mammogram, it shows indications of an illness, but you need more testing to prove it.
2:28 Brazilian GM Leitão (seven-times Brazilian champion) here in Brazil says that at this position the best move is cxb5, meaning, c5 is not a "forced" move at all, instead requires a lot of calculation. What I suspect Mr Regan means by "forced" move is, for example, when I trade Queens you take my queen and I take yours. That's an obvious move, not a "forced" one. What I think Mr Regan does, is to exclude obvious moves from his statistics. These considerations are just irrelevant to whether or not Mr Neimann cheated and only have to do with Mr Regan's method itself. As to Mr Rousis's method of cheating we can only conclude that he is a "dumb" cheater. A "smart" cheater wouldn't commit the same mistake that Mr Rousis did. He would oscilate his performance the same way Mr Regan showed in his statistics to be "normal". So, the Fair Play Commission didn't confirm that Mr Rousis was cheating due to Mr Regan's "excellent insights" (although they eventually were proven right) but due to the fact that they found a phone in the toilet that Mr. Rousis had had installed. 8:45 Here it is stated that one player's opponents were playing the best possible chess while Mr Niemann's opponents were playing very bad chess. So, I ask myself, if such unpredictable counter-statistical performances can happen in chess, how on Earth can I trust statistics to make final decisions that so deeply affects a player's career? My conclusion is, a "smart" cheater, as opposed to what was stated in the video, is a player that cheats with "systematic consistency" but at the same time makes sure he closely adheres to Mr Regan's expected statistics. This means that Mr Regan's conclusion that "there's no reason to suspect that Mr Niemann is cheating" should have a caveat: "unless he is a "smart" cheater and closely sticks to my statistical patterns of "normality".
The highest rated Brazilian Sunday night TV show "Fantástico" just aired a thorough report on the Niemann saga. Check: th-cam.com/video/iCSy96amvMk/w-d-xo.html
It's Magnus who statistically played multiple SDs below his rating against Hans. Statistically, it looks more like he intentionally threw the game, than that Hans cheated.
Normality is defined as what other chess players at that rating would play... And comparing that vs what the engines says. Ken has no input on what IS a forced move... GMs and Engines define that by their play alone. Also, it's not black and white, some moves are critical and hard to see (the ones a "smart cheater" would cheat at), and others are forced... ie, most GMs would agree and the engine also... and there's a whole spectrum in between... Like when 2 moves are about equal, and most GMs find both... Or there's an only move that nobody finds but the engine does... Etc. That's what Kens method corrects for... Which is exactly what you guys are arguing for... and you are correct in arguing for it... it's a sensible thing to argue for... but you are missunderstanding and understimating Ken's work since that's exactly what his method solves. Ken's method would find (and does find) this supposedly undetectable "smart cheater" because it can decide which moves were critical to cheat at, and how hard it was to go right and it corrects for it. and over the long run you'd be found. The fact that GMs think that they would get away with "cheating smartly", just tells you that they aren't that good at stats...
Hang on. Havanna isn't the most suspicious tournament even according to the Regan's list. The first is Charlotte's CCCSAFall. Any chance to see that detailed analysis? BTW that's when HN got his last GM norm.
I work with statistics too. This methodologý is good for amateur cheaters. But the main issue is that a smart cheater would need more games to spot the help, he can even use the help sporadically and even lose the games when using help to show some partial brilliancy. It is the common problem of statistical power.
But if you do that, you don't benefit from cheating. Niemann has a perfect bell curve... Like, as non cheated stats (in the analysed games) as you could hope to get. It's actually funny.
This only looks at the moves so it doesn't look like it even cares whether or not a player wins or losses. Also I think a player making a few 'brilliant' moves but still losing would be suspicious in itself. To be assured to lose they would also have to make sure to make a few poor moves to let the opponent win. When someone in chess is playing well it's unlikely that they're going to make a few brilliant moves, but then also make the kind of blunders necessary to assure they lose. If someone plays a game really well, making some very good moves, they will rarely lose unless they happen to go up against someone who simply plays an even better game. Which is something you can't really plan for.
@@azlastor But isn't his ROI calculated based on his current rating. As his rating improves, the bell curve moves to the right resulting in values that might indicate he is not cheating. Eg If I cheated and moved my rating from 2000 to 2300, the ROI is calculated against my latest rating namely 2300 when in fact I am just a 2000 player)
Okay good addition. Reading this comment it just bothers me that so many people seem to make incorrect assumptions about the model to indicate that he was cheating, I mean statisticallly (lol) many people are making conclusions based on intuition here, and that’s a problem with how most people understand statistics as I’m sure you know. I have a few short questions - Is the claim that “saying that Hans opponents just played worse in a nonsensical, since if he did engine moves they’d naturally play worse” something you’d agree with, or do you think there’s issues with this portion of the model besides that (to me as a layman this ignored the example of the player that played well but played against other people that played well, therefore lost rating, since the methodology for this part of the model wasn’t delved into) - does the fact that Hans Neumann knows how knights move prove that he’s cheating?
@@smartfck4 but he had a good point: we know when Hans cheated, he admitted it, run the Regan analysis for this period of time and see will he get caught?
What isn’t taken into consideration is the massive deviation between how poorly opponents played vs Hans vs how they played vs other opponents. The caster here simply stated “they didn’t play as well” as if it was just luck. If you’re playing the best moves at the right time, it will force your opponent to be innacruate. That number being so much higher should be a part of the analysis. For example, I bet in the end game vs magnus during the most famous game, Magnus had a few inaccuracies…whereas he probably had little to no end game inaccuracies vs others. Why is that? Luck? No. It’s bc perfect play causes innacuracies. Think about it
That would be visible in the roi... the roi was normal so he played according to his rating.. so he didn't cheat and was indeed lucky the others played bad against him. Analysis of the game against magnus also shows that magnus didn't play very well.. it is literally just a world champion who is buttsore he lost
That's an illusion. Stockfish is more likely to exploit your mistakes, but you're making mistakes even when you play a beginner. It doesn't matter who you're playing. On each move, you make a choice. That choice either aligns with a top-engine move or it doesn't.
I'd be interested in analysis of why people tend to play badly against Hans. I imagine it probably has something to do with his playing style and maybe even psyching out his opponents, but I haven't seen anyone go in depth about that
In my opinion is the same motive why his results are inconsistent. He is aggressive and tries to push the game out of the opponents preparation into unknown and complicated territory. That certainly makes the opponents play worse, but it may work against him if the opponent calculates better than him in a specific game.
Well there are players who are known to have these kinds of qualities such as Tal or Shirov, but a grandmaster friend (and former US champion) told me that his opinion of Hans's style was that it was raw and unpolished, and that was probably irritating to more established players.
Are you new to the chess scene? Hans has been losing badly to the top players on all time controls until these last 2 tournaments. It really is an overnight improvement in performance
@@albertopulido4267 don't be an idiot. Magnus has shown why he is on top for over 10 years even since before regular engine use was a thing. He was the youngest gm at the time and drew against former world chess champion kasparov at like 10 years old. Everyone could see even at a young age that magnus was going to be a top player. You make yourself look ridiculous comparing him to a known cheater who up until 5 minutes ago was known more for his online personality than his chess ability
‘Playing style’ & ‘psych’ lol. Talk about a stretch / reach. Cause these super GMs who have seen almost everything crumble upon the slightest deviation or are totally unable to analyze some different situation. ‘Psyching out’. Yea, cause a guy who has fought through the pressure of 5 world championships gets all out of sorts cause of the supreme death stare of this amazeballs 2600 elo 19yr old who has won everything under the sun (o wait).
Regarding the ROI argument (partially addressed at the end), has it been considered that perhaps Niemann has so many performances “below his rating” on purpose? This could be a designed attempt to pass these statistical tests undetected. For example, most of the time Niemann could strategically underperform a bit in “unimportant” events, and when he gets an important event in sight, he could cheat in a cluster of events to qualify and then win the one he was aiming for. I think I saw a different statistical analysis that did show clusters of outstanding performance in time for Niemann, but I don’t remember well. One of the arguments Niemann has made for his recent rise is that he has been playing many games and tournaments, but this could also be part of a strategy to beat the long-term statistical analysis. More games = more chances to compensate for outliers. Thoughts?
The algorithm should be if innacuracies are played at the beggining when the position gets more complex if better decisions are made, it should raise suspicion at the very least
Apart from all the drama, this is super interesting! I wonder if ROIs show different playing style as well (for example a player who plays more non-sound moves but invites the opponents to make mistakes and thus gains a relatively high rating for their accuracy levels vs. someone who plays very solid, a lot of best moves, but draws a lot).
Yes there are other factors as well, such as Hans saying that he happened to be prepared for Magnus's line by looking at a game between Magnus and So. But that game in question was a blitz game, and the idea that Hans would try and memorize the line 14 moves deep from a blitz game before a tournament is suspect (if this were a match against Magnus it would not be suspect). Then, he kind of fumbled some onboard analysis between his game with Firouzja, I believe, and he refused to go into analysis. Right after the win against Magnus, he also refused to discuss the game, saying "chess speaks for itself". You'd think Hans would be excited to speak about his great victory. Most players love to go into variations and discuss the game they won. I saw Ivanchuk crank out variations in his head when being interviewed over a couple games. Yet Hans say, "chess speaks for itself". Apparently not, Hans.
@@leftjedi It's not just a little suspicious. It's like coming home to you're wife buttoning up her clothes looking extremely nervous with a unknown pickup truck parked outside and a history of cheating.
Being a zoomer and having a different personality to chess greats 20 years older than you isn’t exactly cause for suspicion. Got to remember chess speaks for itself line was before anyone talked about cheating, and he was probably just having a laugh and being an edgy teenager.
@@davikarlsigursson5657 this would make sense if it WAS LIKE that, but it was NOT. two different scenarios. This is called the straw man attack, you build a totally different scenario and argue it, and of course it is waaaaay worse, or at least very subjective, but NOT same. He surprised everyone, including Magnus, a notorious sore loser, by beating his ass and making him look very bad, and so Magnus also no builds a strawman attack to remove his embarrassment, and now instead of the world talking about how BAD Magnus got beat by a lower player, we are now talking about the STRAWMAN, how Hans is a cheater. Hans won the battle but Magnus is a master strategist and he is still fighting the battle. In the end I believe we are seeing the END of Magnus reign, as his worse fear is here, a player, a young player who has his number and prepared, created himself as the person who will dethrone the king. All hail the new king, Hans Nieman.
@@normanlove222 he was a nobody who cheated and didn't have the best rating but yet he sacked the best player to ever play. You don't just come out of nowhere and do that especially when you have history of streaming and you getting destroyed. Including the beach games against Magnus which Magnus destroyed him. There's pictures to prove it and a GM that watched it. Enough is enough.
I mean they should step up the anti cheat methods after this drama, i dont see how its difficult to make it impossible to cheat when playing face to face, search the players, scan them whatever, its rediculous that there is a possibility to cheat in chess.
Exactly! and: " 10:16 I absolutely disagree! and think you are biased. It would absolutely give you the result you want: More rating, more money and fame and without the risk of being caught. It's the move of the "very smart cheater". You don't have to win the tournament to improve your financial state. As long as you have a better rating than you could have had - then you are improved. "
I'd like to know more about how Regan's method has been previously used. Rousis' case is a prime example of it's success, the analysis flagged him as suspicious which was later confirmed by physical proof, but how often does a player shows suspicious results and no cheating can be proven? And how often does a player cheat and isn't detected (yeah, I know, this one is pretty hard to get data on). In a perfect world you'd get 100 GMs to play a mock-tournament and have 50 of them cheat so you can validate the algorithm against a known reality.
I would like for that event to happen in public, and for the spectators to be part of the show: they win something if they correctly guess who the cheaters are. Probably a bad idea, though.
The problem is that regans method is and has been fully known and disclosed for over 15 years. All a cheater has to do is a build a system around the known parameters of his anti-cheating method. Regans statistics are very good, it’s just that current method is outdated
Yosha came to the opposite conclution, and said there was 1 one to 77600 for his good results in 6 tournaments on a row, and he had at least 5 games with 100% of his moves equal to the computer.
@@bjorn2fly everything was wrong pretty much. If you do the math correctly on the ROIs she was showing it comes to about 23 in 10000 which ~1 in 500 chance of him having those tournaments (very possible if he recently improved due to a covid break). The 100% correlation she was showing was with any of top engine moves (could be from any engine, so each move from a different engine) which many grandmasters including Hikaru, magnus, etc have had. I am honestly very sus of Hans, but the analysis was useless
Nice follow up! This should work VERY well, most effectively against the “smart cheater”; albeit over hundreds of games necessary for data. Even just one critical move every few games would still show up as a significant deviation in ROI over time.
EXACTLY, congratulations for understanding that! Also, IMPORTANT! HANS data is an almost PERFECT bell curve.. It's not a case of "this data is funny but not skewed enough to call him a cheater..." NO... It's as non cheated a dataset as you could get, that's why Ken is so confident that over the games analysed Hans did not cheat.
This is an explanation that chess needs. No biased arguments, no hasty judgments, just clear, concise and transparent information. Congratulations. Faced with the scandalous and the fanatics on one side and the other, what must be opposed is the concrete data.
First if all I'd loved to see a different sorting being applied to the ROI statistics as it was unclear if there is any kind of time-based trend recognisable. Secondly, for this very specific case, wouldn't it be a valid approach to take all of his opponents, too, and undergo the same analysis in order to have a clear indication if their ROI was just completely off that day and how this compares to their former performances.
Ok, at a certain point, would we be more impressed with Hans if he figured out how to beat the statistical cheating detection algorithim, and cheat OTB without detection time after time than if he had just beaten Magnus of his own volition?
It wouldnt be that impressive tho, honestly if he gets caught by reviewing his previous games he just messed up and overcheated. A GM should be able to cheat a couple of moves per game and not get caught. In his case, some of his entire games have been very precise engine lines. It would be impossible to catch a "smart" cheater that only uses the engine moves sporadically. If we do catch cheaters without actually catching the cheating act (OTB device or an engine running online), well, they were pretty damn stupid in the first place, that's the scary part about the future of chess, people just might boost their performances occasionally and there will be no way to know
Garam - the main point of the video is finding people who cheat on only a few moves. All the people who detect high level cheating know about this and factor it in to their analyses.
hi, i want to share 3 points i was thinking about 1- beside the forcing moves that shouldnt be included in the ROI there is also - moves due to intuition - moves due to prep (not only opening moves) some positions are studied using an engine to the 24th move. if the opponent fell into your prep and you're blitzing moves like hikaru in the condidate tournament then he will tend to perform worse while you're performing better - psychological moves .if you know your opponent is - under time pressure - under pressure of performing well in a tournament to get access to condidates for exemple or the next tournament ( in the last condidate tournament caruana couldl have drew certain games but decided to go for risky plays because he was behind in points) all those 3 type of moves could influence your opponent roi and yours and we hear of them too many times by top players, so they should be taken into consideration 2- cheating isnt just about finding the best moves. just knowing the eval of the position can tell you if the sacrifice made by your opponent is worth it or not. for exemple alierieza avoided a loss against hans by trusting the sacrifice hans made ( obviously thinking hans is using stockfish so if he sacrificed a piece that means there is gains in the position) 3- if the player is just improving, wouldnt that affect the roi
@@statictech7 The tool they are using to analyze his games doesn't care whether it is online or OTB. So comparing the output from his cheating period to the current period would yield some useful information. Much more useful than comparing it with some other known cheater as they did in the video imho. And depending what the results of the analysis are, it could either help his case or hinder it. Either way it would be interesting to see the result
@@fnjfrancis There is a reason to run the test. It's called a control test. We know Han's cheated then, so to test this algorithm, it should also yield a "positive" test.
Many engine "top line" moves are unfathomable to even GMs. The analysis goes to 20 moves or more depth, and is obscured even more by the neural net. So even if a player sees a top-line engine move he may still not understand the engine "plan" for that move and so playing it would be counterproductive.
If you only play one, perhaps, but if you play 5 or 6 moves (even alternatevily) in a match after classic openings, the thing changes quiete a lot. Besides the computer thinks again over your mistakes as its new position. Just play with the help of the computer with a friend, having yourself 5 options to receive help and he not 1, and you will see how you beat him quite more easily at the end of the day. it will be not counterproductive I say. Besides, when you know that is really a good move and you are a GMs is still more profitable, so you can understand it easier and quicker if it is humanly possible.
@@Buymecheap If your answer is to get cheating help for many moves then you are negating the starting premise that states that all that is required is perhaps two or three moves.
Based on what you said at the end of this video, am I following you correctly that Regan's method won't catch someone who only cheats in very limited circumstances? Someone who is mostly playing honestly but cheats on a specific move when they perceive a particularly high payoff?
Exactly! and: " 10:16 I absolutely disagree! and think you are biased. It would absolutely give you the result you want: More rating, more money and fame and without the risk of being caught. It's the move of the "very smart cheater". You don't have to win the tournament to improve your financial state. As long as you have a better rating than you could have had - then you are improved. "
Kudos to you for pointing out what so many of the other videos have been missing: you always have to look at the quality of the opponent's play as well.
Yes, if Nieman has a magical psychological ability to make his opponents under-perform (maybe because they were under-estimated him. "oh today I play with this kid with the funny curly hair?") , then it is something that can help his results.
Perhaps the suspicion of his cheating caused his opponents to play worse. In the game in which Magnus accused him, Magnus played well below his usual standard. He was clearly distracted by the cheating issue.
Your conclusion "Statistically there is good reason to believe that he is playing honestly", although carefully worded, is not correct. The analyses done by Ken Regan, and he would undoubtedly agree, only lead to the conclusion that cheating is not proven, not statistically, but that only according to the particular statistical methods used. That is all. This conclusion is entirely dependent on the particular statistical analyses used. There could indeed be other statistical analyses leading to the opposite conclusion. Nobody can exclude this, and certainly not a serious statistician. To conclude as you do, one would have to establish that this is impossible... On the other hand, the same kind of reasoning leading to your conclusion would lead to the conclusion that before 2012 there was "good reason to believe" that the Higgs boson did not exist! No, it just hadn't been detected. HN could be a kind of Higgs boson... :-) Finally, knowing the statistical methods used by Ken Regan, it is possible to cheat without being detected by these analyses: it is enough to couple the chess program used with a program ensuring that the statistical distributions will be consistent with those expected. This argument alone is also sufficient to refute the possibility of reaching your conclusion.
That's just not an accurate way of checking if someone is cheating. I actually thought that Ken Regan would be going straight to the source. Cheating in modern times is almost always about using a chess engine to give you your next best move. As such I would expect that an anti cheating algorithm would connect directly with the top chess engines out there (ie Stockfish and Alpha zero) and then correlate every move with the engines' best moves (while also ignoring openings and forced moves) and then yield a correlation percentage based on how many moves match these engine's top moves. That's what made sense to me. Instead, Ken Regan is using an engine to tell him if a player is cheating based on his assigned rating? Hans has been cheating since he was 12, which means that you simply have no idea if his rating should be trusted in the first place. As a thought experiment, if you were to lower Hans ranting, by deducting 200 points his assigned rating at any given point in time, and run the analysis again, how many of his underperforming games end up above 50 ROI?
Your straightforward way of directly comparing with engine moves would probably have a lot of super GMs branded as cheaters, based on where you put your cutoff value. And trash players like me could cheat a whole lot without being found out. The problem of the long-term cheater is pretty interesting. But first, if Niemann, like you put it, "has been cheating since he was 12", (semantically not wrong, but pragmatically implying that he's been cheating pretty consistently all the time) how do you explain that his extreme rise in elo is a lot more recent than that?
You said that with Regan's method, only a very limited number of moves per game are relevant for the analysis and that you you could cheat on some(none relevant) moves without being detected. So that makes me wonder, if it is possible to figure out which kind of moves get ignored by the analysis and systematically cheat on those moves. Because as other people's analysis shows, that there are indeed factors in which Niemann clearly and consistently deviates from the majority of other (most notably stronger) players.
yeah but why would you want to cheat in a position where there are like 8 moves giving you e.g. a engine evaluation starting from +0.4 down to +0.2. If you want to cheat your way up cheating in the important moments is what you need.
The moves that are ignored are typically moves in which cheating would not matter e.g. forced moves, opening theory. The only possible one is when there are multiple winning variations, which, as said in the video, it is not uncommon for a highly rated GM to find one.
The moves that are ignored are ignored for a reason. If someone makes moves that make a significant difference in the game, those moves aren't going to be ignored anymore.
I am more confused than ever. There was a recent analysis of Hans' games and it showed that Hans had 100% accuracy in TEN games on chessbase in OTB games. No other player in the database even has one. One player got a 98% and was caught cheating. How does Regan arrive at the opposite conclusion?
@cDb lol, yeah. As a GM I would figure that he would be aware enough not to send up red flags left and right. Who knows what this "correlation score" is and other people have said they ran a correlation for those games and they weren't 100%. Statistics are so easily manipulated.
That's wrong, many players have 100% games. The comparisons made in Yoshas video are over much longer time periods or at least whole tournaments, comparing them to a single game is unfair. The comparison in itself is unfair too, because Hans was playing people rated much lower than his skill level (if we assume he is legit) so achieving a high accuracy is much easier compared to when super gms are playing against other super gms, which is the reason why for example Hans never had and will never have a 100% or likely even a 95 or 90%+ game against any of those super gms. But against 2200-2400 rated players it actually isn't that weird and GMs like Caruana looked at the games by themselfs and found most of the moves to be rather human and natural.
@@LunnarisLP many players we are talking about here - GMs - are lucky to even have a 90+ in their entires careers. “many players have 100% games…” LMAOOOOOO 🤣🤣🤣
FYI, Hans is a known cheater, just not OtB. I don't agree with the argument that the motive for cheating is to get a high placing or to win for the financial incentive. Hans had not until recently been invited to any major tournaments. All he wants is to continue to make these large tournaments. He is after the fame, and as a content creator, the financial gain that comes from it. All he needs is the occassional win against top players. Good luck catching that with statistical analysis. In his game against Magnus in the Sinquefield Cup, after the opening, Hans made all top 3 moves (according to stockfish) and on three occassions made the only winning move. That paired with his refusal to explain lines and variations in the post match interview is incredibly sus. With a history of cheating, his quick rise in rating, and considering his age he should be scrutinized.
the big question though is how the hell can he cheat in OTB? OTB anti-cheating regulations have been extra strict since the Toiletgate scandal between Kramnik and Topalov... some players have actually complained about that... Hans must be using CIA-type technology to cheat in order to evade those anti-cheating checks
Ha ha even M Tal does not know all the lines when he made sacrifices and he admitted that. Sometimes you can't explained all the lines when you made a move, just relying on intuition and feeling for the position. Trying to dig all the lines and variations during the game will lead you to time trouble in most cases. And what would you expect to Hans when it comes to media interviews ... he is very young and less experience in responding to top level media interviews that's why he can't explained many things clearly and made some erratic statements.
@@excalibur92 according to your logic, even tal doesn't play 100% accuracy. This hans guy won with 100% accuracy more times than Tal, Bobby, Magnus, Morphey combined. Let that sink in, trust me, he's not that great.
@@excalibur92 Tal and his opponents didn't have access to strong chess engines that are available today. His playstyle was based on previous chess knowledge and intuition. And there is also the intimidation factor. His opponents might have just assumed that they are worse after he played some sort of a sacrifice and they might have lost their concentration. When it comes to Hans's interviews, not only did he not know how to explain his moves but he also said that his position was clearly better when it wasn't clear at all. Even Hikaru Nakamura admitted that during his livestream. You have the right to be suspicious of somebody who has cheated in the past, was banned from chesscom and whose coach is also a notorious cheater.
@@maksymiliank5135 Many top GMs today when they played a move and ask to explain it after the game ... you can always hear from them the Russian expression "I don't know" when it comes to evaluating and analysing their games. Then maybe they can also be accused of cheating?
This is really really well done. Pleasantly delivered, calm, sober, friendly, straightforward, knowledgeable. It's a pity we're getting this only by now. Congratulations.
@@zaimnaqvi8893 " 10:16 I absolutely disagree! and think you are biased. It would absolutely give you the result you want: More rating, more money and fame and without the risk of being caught. It's the move of the "not a complete idiot cheater". You don't have to win the tournament to improve your financial state. As long as you have a better rating than you could have had - then you are improved. "
@@higamato3811 What about the argument, that he could have cheated only sometimes, don't you agree with? This way he wouldn't get caught by this statistic analysis and also improve his position
Id love to see an analysis of critical moves. I think measuring cheating with all moves included (except forcing moves) is still too broad of a scope. A smart cheater would only cheat at pivotal moments. (Such as moves that help you survive an attack or brilliant moves that will win you pieces) I'm not sure how those could be measured but it would require much more rigorous analysis of each game.
@@excalibur92 yeah and then in 10:10 he admits a smart cheater wouldn't cheat in every game and as a result would never get caught. Regan concludes Niemann didn't cheat, despite Niemann having games 50-60 on half that list like Rausis, the ones where he didn't cheat are used as....evidence he never cheated. The assumption that a cheater wouldn't benefit by cheating only like once in a while in the video is so wrong just by this very controversy and how Niemann's rating jumped to 2700 after one victory over Carlsen
@@jacobpeters5458 I’m not sure how those two things are connected (spoiler: they’re not). It’s a bell curve. If you didn’t have ~half your observations above 50, then it wouldn’t be normal.
@@jacobpeters5458 No, the video explained his opponents frequently played worse than usual, but only against him. I would wonder if there's some correlation between cheating rumors and his opponents beginning to play poorly versus him, but you would need a source in GM circles who remembers when the rumors first began circulating. Also, one could just analyze the games and see, but then you are doubling up on work Regan has already done.
What does Ken Regan algorithm say about the games Hans played earlier, during the period he admitted cheating? It's obvious that he cheated more than those two times he was caught. Algorithm should be ringing alarm bells.
So according to many GMs, when you're playing against someone who's suspected of cheating, your performance may drop because you can't focus that well. I believe this is exactly how Hans cheats - he spreads rumours about him being a cheater, e.g. during the infamous interview, in order to stress his opponents and sometimes even get free wins, as he did with Magnus. It's ingenious, really.
This is the flimsiest thing I have ever seen on this cheating topic. By your metric, I can say Magnus’ title as the best in the world is what makes other people play worse. The pressure, and the stress of having to perform. In fact my example is far more likely than yours.
@@anatolydyatlov963 I was 50/50 on whether to is was sarcasm or not, because this wasn’t too far from what a lot of the chess community would actually speculate lol. So I decided to bite, the last three words would have confirmed the joke but I was at work when I saw the comment and didn’t read it entirely. RIP.
Centipawn model is far superior to this when it comes to analyzing expected rating vs. actual rating and there it is clear cut that Hans cheated. Why come up with a much more complex model when the more basic stuff is clearly superior? As a Mathematician myself I can answer this: it is because the simple stuff can be covered in one published paper and that's it, there is no milking that cow. Coming up with a more complex model even if it is worse, will allow you to milk that cow over and over by further refining it which will subsequently allow you to publish more peer reviewed papers. If anyone doesn't know: the quality of a professor at a university is usually measured in the number of papers per year, not in the quality of said papers, this gives a huge incentivo for people to publish trash research just to increase their productivity instead of publishing great stuff. Also as a tip, if your research is good, your academic credentials don't matter (Einstein was a post office boy), so starting the discussion with "this guy is a PHD of X, from the University of Y" this is a huge red flag of someone trying to appeal to authority BEFORE presenting their argument which could indicate dishonesty.
Ridiculous. If cheater knows algorithm that is used, he can easily make the strategy. There is also random cheating approach. So, total BS. Not saying that Niemann cheated in the over the board tournaments. But his history of cheating online would always haunt him. Him admitting cheating twice and only when he was called speaks volume too.
Easily make the strategy? No, he couldn't. Must be nice to have such a big ego that you think you know better than the foremost expert on cheating. Admitting he cheated speaks volumes about the fact that he has changed his ways. The fact that he cheated online when he was literally a minor was well known in the chess world long before he spoke about it publicly. No one would volunteer that publicly if it wasn't necessary.
I'm not a high rated chess player, but it seems to me that when I watch Hans' games he always seem to throw in a few curveballs in most of his games. He doesn't play the straight forward theoretical moves all game and I think this throws off his opponents. This is not to say he's a better player or faster at calculating lines but rather he's confusing or tricking the opponent. These moves he plays are not the optimal moves so they can give the opponent advantage if they calculate a reply properly. At first I thought he was probably cheating because of the way he so easily beat Magnus, but the more I see from him the more I'm convinced he probably wasn't cheating. He's just figured out a different way to play and right now he's got the advantage before other figure it out and begin to counter his play. I really hope it comes out that he isn't cheating because it'll make chess even more interesting. Really want to see how long he can keep up these great performances before or if anyone can finally figure him out.
if i will play against an engine it is obvious that my accuracy will drop, u cant say that his opponents played badly against him in order to prove his innocence.
Exactly, and: " 10:16 I absolutely disagree! and think you are biased. It would absolutely give you the result you want: More rating, more money and fame and without the risk of being caught. It's the move of the "very smart cheater". You don't have to win the tournament to improve your financial state. As long as you have a better rating than you could have had - then you are improved. "
I’m glad this video was made to explain Reagan’s results. I am not personally satisfied with the rigor of the analysis. Perhaps there are more metrics that are not being discussed, but there should be more than one method of determining whether or not someone is cheating. Caruana suggested Reagan apply his methods to the period of time that Hans admitted he was cheating. Caruana suspects he would not detect anything, hence the need for those with the most to lose to demand better standards. Modern times call for more creative solutions. Otherwise chess may fade into the black.
@@JosBroder Dude i thought you were talking about a proper analysis. She has been debunked hard. I'm not trying to blow you off or exit the argument. It's just a really sad hit piece by her and it's been discussed a lot at reddit etc..
@@JosBroder Good on you, glad people are keeping open minded here. The main 2 issues were: The engines vary depending on the player. Hans has hundreds of engines used for his moves, whereas Carlsen and everyone else only has a few. That is because of the way that system works (explained in more detail elsewhere). 2nd issue is she multiplied raw odds in a very amateur way. For example if he played 3 great moves in a row she would say the chance is 0.1 * 0.1 * 0.1 = 0.001% . But you can't do it that way.
I was interested in Regan method since Patrycja Waszczuk case, thanks for explaining. Regan missed important detail in my opinion, opponents, human play poorly when thrown against a computer forced variations, this could be the case of Hans opponents 0,17 accuracy...
Excellent video. Have just arrived at this video after getting back in to chess at christmas, and I'm glad I did. Makes me see this Hans Niemann fellow in a much better light than most of the videos suggest. He could be very hard done by. In either case, the best way to prove his innocence is to realize that there are a lot of people on his side, perk up, and keep playing well
This was a very good and much needed video. When ken was explaining I had no idea what on earth he was talking about. But you explained it perfectly for me to understand. All that’s left is to see what magnus says in the coming days
In my opinion, the table at 6:15 is the key. If all Niemann tournaments sort by date, it is possible to look, if his ROI is strong correlated / dependent with date. If yes, it means, that Hans starts significantly cheating in specific moment. But if there is only weak correlation (or none), then it looks to be OK. I dont say, if Hans is cheater or not. I only suggest the statistic experiment, which could better explain / uncover this unclear situation.
The ROI analysis is pretty well useless unless wild assumptions are made regarding the nature of the cheating. Realistically, you would ONLY see ROI discrepancies like Rausis's if the cheating was consistent. Its pretty clear that pretty much all players agree that Neimann doesnt cheat every match, or maybe even most matches - he frequently performs quite poorly at tournaments. If you cheat very infrequently, then it would actually be MORE likely for the ROI to show you as playing "under your expected rating" if you are playing a typical match without assistance. At the end of the day, its not really strong evidence either way. Theres just way too many unknowns to draw any real conclusions. The only thing this analysis shows is that Hans didnt cheat the exact same way as Rausis.
Very good stuff. But personally, based on what I saw, Magnus had prepared an opening that was unique in his mind. But Hans said in post interview that he had been preparing for that particular opening. Wondering if he had a mole in the Magnus camp. But would that be cheating or just 'research'?
In the next days interview, Hans clarified he studied a transposition, which did end in the same position. So not the same opening, but the same position reached from a different opening.
Very good conclusion summarising what really happened here. And the details to back it up. This statistical analysis only exposes certain types of cheaters. And it's very easy for a wise cheater to circumvent. THIS METHOD CANNOT PROVE SOMEONE ISN'T CHEATING. A lot of people are missing that point and misleadingly summarising the findings.
I think trying to give reason for cheating is flawed. Cheating isn't rational, especially if he cheated specifically to try and impress Magnus for one game. Sometimes cheating is used for consistency or monetary gain, but as you said it would be much easier to spot if they used cheats that way. I'm sure a cheater would try and look deep into cheat detection to try and side step it. Not saying he did or didn't cheat, but I don't think we have enough to get an answer.
If there is no reason nor any objective the cheating results are mostly irrelevant and can be ignored. If a person cheats in a single games on a whim it is certainly wrong, but it does not impact himself or anybody else in a meaningful way and it can be ignored. The cheating that must be caught to allow the game to keep fair to everybody is consistent cheating that alters rankings and important events results. This won't be whimsical cheating.
@@conelord1984 interesting point, but you could argue he did it for a single game to get attention for example beating Magnus. But cheating is relevant to who he us unfairly beating even if it doesn't effect his overall rating!
@@Flawlesslaughter If he did it in a single game to get attention on this game he got a lot of unwanted attention. Overall it is not very important though except for the bad publicity for chess and the blame for that was mostly Magnus and not his even if he had cheated (which I doubt). It is easy to discard this game in specific for other reasons though. It was not Hans who played to well, Magnus played horribly. He was in a bad day and that is it. Additionally Hans is not being suspected of cheating in that game only though, especially by Magnus. Magnus was already hostile to him before this game ansd made hints about his quick ELO climbing.
You've removed the first conclusion, which was that Niemann is progressing in an unusual way, therefore there is a reason for it. No reason has been discovered why Niemann is moving upwards oddly - it may not be cheating - but that is what to search for; not just "evidence of cheating."
That’s just not true… Hikaru destroyed him last week effortlessly, and made the difference between a high Grandmaster and junior Grandmaster quite evident. The only reason high Grandmasters would be afraid of him is if he was cheating, which seems to be the case, at least historically. With Hans self-admittedly cheating previously along with his coach, can you blame top tier masters of the game being ultra-suspicious of Neimann and his inconsistent play?
8:12 - a lot of people have already pointed it out that opponents just suspect Niemann of cheating and therefore perform bad, it's not Niemann playing like Stockfish 15
He won't do that because Ken Regan's method of analysis is kind of a joke. There are players who have been caught red-handed cheating and his analysis doesn't show up as them cheating
As GM Fabiano suggested, Dr. Ken Regan should analyze the games Hans played in which he admitted he cheated. His prediction was that Regan's method would not catch the cheating that was admitted to take place. Frankly, a smart cheater would know not to cheat in every game, on every move. Instead, subtle cheating could be used as both a confidence boost as a safety net. This is why OTB matches should be played with a great deal more scrutiny and security. It's possible that subtle forms of cheating could happen.
It should also be noted that Hans has dozens of games with close to "perfect" according to computer analysis, far above the level of perfection of World Champion players. This is not proof, but considering that Hans was also a late-bloomer, who instead of becoming a GM at 15, he was still cheating at 16 to boost his Chess streaming career.
@@timefororbit Well, two points then. The first is that Dr. Regan did analyze the games Hans admitted he cheated, and Hans came up as having cheated. This is in the report recently released by Chess.com. The second is that the so-called perfect scores are the result of manipulation using Let's Check, except one can maneuver the results up or down accordingly. I have no real desire to do a video on such, but I can assure you I have tactical slugfests by Kasparov, to name but one, with 96% correlation and the like. The thing to remember is that Let's Check is not a single engine. You need only match any of three per move to be counted a match, and you can then supplement them with an engine of your choice.
I don't care about all that science numbers or math. Magnus is saying that Hans is a cheater. That's all I need. Magnus never did this before, so that's it.
And what about when Magnus himself cheats - Here is the full stream. The events in question are at 1:44:00. Lots of talk of cheating and banning. th-cam.com/video/CRdrf1Ny3x8/w-d-xo.html
Ken Regan's analysis reveals something very interesting: it's like Hans can nfluence the (playing) behaviour of his opponents. Maybe Magnus felt this influence very clearly, since his chess-thinking is co clear. His coach could not comment on such a 'vague' thing like that.And Magnus cannot 'prevent' something like that either so he chooses nog to play him. If true this leaves us guessing about how Hans could do that. I cannot resit the thought of 'black magic'.In all sorts of cultures you can go to a 'dukun' (kind of sorcerer) and buy a 'spell' that confuses the thoughts of your 'enemy'. Very hard to trace and to prove.
Good explanation and I think Kens analysis should continue to be used however it’s definitely not sufficient a lot of gaps and the forcing move thing is kind of bs, shows lack of understanding. It assumes its easy to find the forcing moves.
Did you look up how the model defines forcing moves or did you nitpick on instinct just so you can confirm your current opinion and move on? Do you think the model defines forcing moves as obscure engine moves that Super GMs would fail to find that make them slightly less losing or equal with their opponent?
@@sams7068 Actually I did look it up but I haven't seen a paper that details the algorithm, which I suspect is not publicly available anyway, however, I would love to read it as a mathematician. However, using the example of c5 it is a fair assumption that the forcing moves assumes a big deviation between top lines and other options. I don't see how "easy to find" can be defined in mathematical terms as the prior reply suggested thus it must have mathematical basis to it such as big drops in evaluation with one/two options available. In the position shown cxb5 and c5 are the only two moves that don't loose on the spot for white. However, if memory serves me right c5 was played rather quickly, when cxb5 is intuitively the more obvious move. b4/b3 with Bxe1 is a big threat in the position that cxb5 actually avoids. Thus, from a chess perspective and not a statistical perspective its obvious that considering a move such as c5 as forcing is somewhat weird. This could indicate that in an incredibly complicated tactical positions such as this one the one or two options that are not loosing on the spot will always be considered forced moves, when we know for a fact humans are bad in complicated positions and shine in positional chess. I also understand the idea of forcing moves in Ken's model as it is an attempt to avoid false positives, but the result could be false negatives. I also have no bias of opinion, my opinion was formed on the video, as a matter of fact I was extremely interested to learn the information in the presented video and I do find the model in itself to be brilliant conceptually, however, every model has limitations.
@@sams7068 as a matter of fact Igor Rausis never scored particularly high within the model, the give away was the consistency. He got banned and stripped of his GM title because he got caught with a phone in the toilet analysing his position. Up to that moment FIDE did nothing and the investigation only started after he was caught. He might have been under suspicion for ages, but I will tell you for a fact many players were speculating he was cheating before FIDE caught onto it and the guy got away with it for at least 4 years.
I like presentations like this where you present a set of facts and leave it to the audience to make their call rather than some prominent GMs people who make deductions thereby biasing their audience.
This video deserves more views. Nice and interesting analysis. Hopefully more players will get a better understanding of the anticheating methods and also the difference of perfoming high in a tournament or a game based on rating performance vs a more objective approach and analysis of what it really means to perform better when you put the games under the microscope and computer analysis combined with a sophisticated approach of discarding forced moves and opening moves.
The whole idea behind Regan's system depends on knowing the actual playing strength of a player. For non-cheaters, you do know this, it is their rating (with a small to moderate +/- because of form, age etc.) For cheaters, you have no idea of their actual playing strength. So you have a cheating detection system that is only good at catching the people who actually do not cheat. It is still not a useless system, because it can detect someone starting to cheat.
The difference between player’s rating and the computer analysis is that rating only takes into account actual outcomes of the games against other rated players. The analysis looks at the actual moves in every game and compares it to the model corresponding to each player’s strength. Cheater would have to cheat very consistently or at every move of every game to make sure they are not caught. He would have to mix wins, losses and draws, make blunders and also finding winning moves only according to his rating. While this is possible, it is very hard to pull off especially over the longer periods of time.
But HOW do players chess players cheat OTB? I don’t understand the ability for a transmission of next moves. If the game isn’t being streamed, or is but with a delay, how could anyone assist in the cheating?
Mobile phones in the washroom are the obvious culprit, because in smaller tournament, nobody is gonna check what you do in there. But even on GM level, there are various way to cheat, such as your partner in crime might be simply visible to you and giving your coded gestures.
The idea that other players play worse against him rather than him playing better is a nonsensical statement. To play moves that your opponents find difficult to respond to is your better better play, not their worse play.
So with a lot of the recent games by Hans where he is winning in 28/32 moves... And all the opening/forced moves being ignored would there be enough moves to identify anything? Also if a cheater is aware of this method of ROI can they not make sure they have bad days to balance it out?
That's the other thing. By Regan revealing the method of his analysis, an astute cheater can use that to devise a cheating strategy that would fit that model's definition of fairness.
It is move level (or an aggregate of move level, to be precise). Roughly like this: Take out all insignificant moves (forced, opening), calculate how large percentage you find a top move[1] and normalize (using skill level, I suppose ELO here) so that 50 becomes your expected value. I guess the problem is that most super-GMs seems to think (see Hikaru, Caruana, Carlsen) that even having a single binary signal during every game (like, "in this position there is a huge difference between moves and if you find the right, it is basically winning") would make them crush the field. It is in fact quite similar to in poker. Obviously RTA (real-time assistance) is banned from online poker, but if you do it sometimes and selectively, it is basically impossible to detect, I would think. I think finding only the crucial moves, maybe even in crucial games and doing statistics on those (maybe correlated with live/delayed brodacasts and with/without audience) would be interesting. My gut feeling is that this drama will be dissolved with data from Chess.com, though. No OTB cheating proved, but a much larger sample with super likely cheating and thus credibility of Niemann drops dramatically and he'll not be invited "anywhere". Sorry to say, but there is actually only a very small chance that he cheated only exactly when caught, which he sort of claimed in that otherwise seemingly trustworthy interview after the SC... He has also been awfully quiet after Chess.com basically doubled down and said there was more to it than he admitted. All of that being said, super interesting analysis and shows that it is not clear cut and easy to see here. Also easy to cherry-pick data, have not yet seen qualified opinions on this video th-cam.com/video/jfPzUgzrOcQ/w-d-xo.html showing some very high correlations with engine for particular Hans games. [1] - I am not a good chess player (enjoy the game a lot though) and my statistics skills are rusty, but I guess computing this based on centipawn loss would be the straightforward route.
In casual, non-tournament play, people were blowing Hans out of the water. He admits he cheated 4 years ago & 8 years ago. A leopard doesn't change his spots.
Well, a cheater doesn't have to use the top engine move very often. For example in a position where the best move gives evaluation +5.3 the second best move gives evaluation +3.5 and the third best move gives evaluation +2.3 he may as well play the third best move [which will be considrered an error of magnitde 3 (because 5.3-2.3=3) and win as well]. Also in a position with great complexity he may also from time to time select a move that gives evaluation +0.7 and regardless win the game if he opponents fail to play perfectly due to the high complexity of the position. If he uses suboptimal moves that win regularly, being a "smart cheater" then the analysts will be unable to detect with certainty the cheat. I personally believe that Hans Niemann is a cheater. The main reason that I believe that about Hans is that, unlike any other 2600+ player, Hans is unable to provide a consistent analysis of a position in postgame interview providing lines and variations.
Kinda late, but as i understand, playing top engine moves is worth it when you know how to capitalise on them. By that logic, Hans couldn't get away with them, 'cause he can't analyse at the same depth as engines and make following moves that engine thinks would give you the advantage, i guess
Would it be a fallacy? Because if he cheated before, he gained certain points of Elo. Further analysis will be applied on his cheated Elo. So the ROI makes no sense, if he have cheated at some games. Or am I wrong?
it would make his performance way worse unless he was using it all the time - if he wasnt capable of actually reaching grandmaster level it wouldnt really matter if you cheat sometimes for ROI as it would bomb during the games you werent cheating in due to the inflated rating
@@fredericpelloud7536 think of it like this ROI is essentially your average expected performance for your rating on a bell curve with 50 being your rating and 100 being playing so unbelievably far above your rating youre like a 1500 beating a 4000 rated engine. If neimann truly has a lesser level of skill than his opponents, when he doesnt cheat (to avoid suspicion) he would immediately become severely less accurate and his ROI would become suspiciously low. Also online chess (where hans cheated) doesnt actually matter for FIDE rating afaik?
If you only cheat occasionally, you're far more likely to get away with it. That's where Igors went wrong. Also, if you do worse performances elsewhere on purpose, you can cause it to average out and make it look more normal. You're completely ignoring the abnormal peaks.
@@Sarah-oj7bh nobody else has comparable engine stats to his supposed peaks. Literally not a single person in the world. Not even remotely close. It's obviously abnormal.
I don't know--I still trust Magnus's OTB instinct when he felt that in their St. Louis game Hans didn't spend much time concentrating on critical positions, and beat him in an endgame in a way only a few players could (Magnus would know, since he's an endgame wiz). You can't quantify this feeling. But I think the World Champ can "sense" if someone's play is a bit fishy, over-the-board, just like how a strong master disguised as a weak player plays a street hustler, and immediately the street player notices they are not that weak. You just know, based on the moves chosen, and the time taken over the moves, the hesitation, the expressions, and other "human" elements.
@@tajneeley Sure there was! Did you even go over their game? The endgame had all sorts of tricky pitfalls and traps, positions where Magnus could usually squiggle out of, due to his superb endgame maneuvering. But Hans kept playing the best, strongest move each time, pinning Magnus down. That's probably why he got suspicious, as he said. If you think that sounds stupid, then you think Magnus is stupid.
he doesn't have to defend himself, they didn't even provide a single proof against him, that's like they are barking dogs. tell me, would you talk to a dog if he starts barking ?
I'm glad to see an expert concluded that Niemann was not cheating. For those who still think he was cheating, I have to ask: "How did he receive the move information from the engine??"
It's great for catching consistent cheaters. Not so great for catching opportunistic ones. If you pick your moments, and play without cheating a fair bit of the time, you'd have a normal range of variance in performance still. It's not hard to defeat a summary analysis which is looking for averages and a regular range. You really want a model to detect an anomalous move, not a consistent over performance these days. If you play most of the game at your standard level, but then on a crucial move, get insight, and then continue playing at your level, it wouldn't show up on an analysis like this, it's comparing your overall performance.
The problem is that you can always play moves that are "too good" just because you overlooked some other human moves. The only way you can detect such a thing by pure analysis of the game (i.e not catching the guy red-handed) is to see discrepancies on a larger number of matches. The data is showing that Hans's games can't be distinguished from the game of a good player.
@@desarguesbaptiste5577 the data isn't even showing anything here, because it's from individual tournaments which averages out instead of being shown from individual games. Now that's the data I'd be really curious to see since suspicious games would look really anomalous
Great logical deduction! absolutely agree and shocked by the illogical conclusions of this video. How can they not see what you've just said? " 10:16 I absolutely disagree! and think you are biased. It would absolutely give you the result you want: More rating, more money and fame and without the risk of being caught. It's the move of the "very smart cheater". You don't have to win the tournament to improve your financial state. As long as you have a better rating than you could have had - then you are improved. "
If Magnus says he’s cheating then he’s cheating. I would take Magnus’ words who was seating right in front of Niemann for several hours during the live game over some expert who studied his games over brief video clips and computer analysis.
If you are being serious: this is hilarious If you are trolling: you’re not funny World chess champion has somehow found a way to be a master of body language analysis, a field in which the methods are so inconsistent and unreliable that it’s relegated to the arena of pop psychology pseudoscience
Just one thing tho, besting the current World Champion (or any sufficiently renowned top player for that matter) would give quite a reputation. It may give the impression of "an underdog that defeated top 1 on multiple occasions". Metaphorically speaking, it's not about an immortal bleeding, but for a mortal to make an immortal bleed. I still remain neutral at this drama, but there's no doubt this will make Niemann either famous or infamous. If he came on a better light after the drama (Cheating not proven, Carlsen acknowledging him,etc.), then he'll be famous for standing up to all of that pressure, one of them from the World Champion himself. If he came on a bad end of the drama (sufficient enough probability of cheating or something along those lines, but still not enough to ban him from playing) then his infamousness would be a good psychological card. Imagine playing against a probable engine or just a mere human. You would even doubt if they blunders.
I think the obvious conclusion here is that Hans isn't cheating, he's just blackmailing his opponents to play worse. Magnus didn't drop out because he suspected cheating; he dropped out to make sure his family was OK.
Why sort on ROI and not time, what if he plays five tournaments in succession with a ROI of 60, that would be suspicious right? or after how many tournaments would it be suspicious?
1) If your opponent plays very well, is it not expected that the performance of the opponent (Opponent's figure) will automatically be lower? 2) Don't you need a baseline to generate the 'Opponent's Figures'. For example, i) the statistics (Opponents figures) of GMs playing blind against a chess engine and ii) the statistics (Opponents figures) of GMs playing blind against a chess engine where a human is purposely adding fake moves in order to try and confuse chess cheating detection software? 3) Knowing that only certain moves are used in the statistical analysis, means that a chess player can use cheat moves sparingly in the middle game, but use cheating to make perfect opening and endgame moves? 4) Recognizing forcing moves may not be obvious, so cheating to recognize forcing moves would not be counted in the statistics as people assume that every chess player will see the forcing moves without the need to cheat?
Innocent till proven guilty (and everyone benefits from that and you won’t appreciate that unless you are wrongfully accused for something, not just chess). Hans is probably using unorthodox moves or psychologically impacting others in unexpected ways combined with having significant growth in certain aspects of his game (making him erratic to some) as his game improves overall. Think of a teenage tennis player who has worked tremendously on his topsin and slice serves for a few months, improving greatly in those areas (while improving at a more regular rate in his overall skills) and perhaps who has studied the games of a few of the people better than him (i.e. Magnus’ games in Hans’ case). And it’s possible he accidentally or intentionally found out about Magnus’ plans in one game. Has he admittedly cheated in the past online? Yes. Did he cheat more times than admitted? Possibly. We’re also dealing with young teenagers learning about themselves, life and the “philosophical” rules in general. He’s clearly a bit of a rebel aka independent thinker. Give him the benefit of the doubt OR trust and verify. Let’s see how he fares in the coming months and years. And as intelligent/skilled as Magnus is at chess and even though he has been right before, he could be wrong. Many elite and brilliant CEOs, surgeons, pilots, etc. have made catastrophic errors in judgment and have been wrong to the peril or demise of many others.
"Innocent till proven guilty" He admitted to him cheating twice when he was caught. So, guilty when was caught and innocent when not. "He’s clearly a bit of a rebel aka independent thinker." Because of him cheating online? LOL But Niemann cheating (verified in the past) will haunt him.
@@mytube2011z If he is still there playing, that means he was not perm banned. So we have to live with that. Innocent until proven guilty for now. If that isn't enough, they should make new rules to prevent anyone who has any historic of cheating for now on.
Han’s opponents in crucial games may have been flawed by his speed of calculation during ambiguous positions. A deeper insight into his facial expressions throughout questionable games may shed some light.
What do the odds in Regan's models say about Magnus misjudging a known cheater having just played actual matches with him? All we have to do is input all the other times Carlsen has resigned a match, quit a tourney, and publicly commented with an accusation, and... let's just put that data into the formula and... oh, wait...
So we have a problem now, as there has been shown that Hans has played more than 20 games with 100% engine correlation. Magnus has palyed 2 of these games in his whole career, Bobby Fischer not a singe one and his engine correlation was 84% during his winning streak. Basically there`s strong statistical evidence now, that Hans has used an engine to cheat in chess. Seems like this analysis here is wrong. Time to reevaluate the method.
It seems like only a matter of time before its proven. Thats the problem with cheating in the modern age. Every game is logged and can be analyzed and compared. People say "there is no evidence". But I would say, if this were a court of law and you presented a jury with his engine correlation statistics they would probably deliver a guilty verdict. What possible defense is there for that? He just randomly morphs into the greatest chess player of all time?
i dont agree with this theory, can a cheater become better at cheating so that he wont get caught ? yes ! notice as his ROI improves so does his opponent, this means he is adjusting the level of cheating according his opponent. He is smart not to go full Stockfish that would be pretty obvious, he is a prepared cheater and throws in random moves ignoring the feedback from Stockfish sometimes. Intrestingly this is EXACTLY what chesscom had released of Niemann's online chess games and their Antii-Cheat detection, which isnt fully expĺained how that exactly works but is does describe a part of it's functionallity that it reckognizes someone using every available attack or defense from the chess engine. As some experts pointed out that not even the best chess player in the world can memorize every strategy that was ever written in a book. And Hans Niemann when asked about his moves didnt gave any correct anwer besides his incoherent rambling.
So if you're comparing move accuracy to rating, how do you tell the difference between a cheater and an underrated player? Or the difference between a consistent cheater and an appropriately rated player. This analysis seems to be asking the question "is a player making moves consistent with their current rating?" In the long run, the answer to this should be "yes" by definition, regardless of whether the player is cheating or not.
Indeed - it also completely ignores the fact that a competent cheater would make an active effort to smooth his results when graphed. It would be insane to make the best move, all the time, in every game, and come in #1. A person's goal may be just to slowly raise his rating to get into the higher tournaments and then to take a good payout and move on with life.
I think there are things that come along with an underrated player. Let's look at niemann for example, since his chesscom account is open, which I assume is after his previous ban, he played more than 2500 bullet games and 1000+ blitz games. Für bullet it's already really hard to cheat, but overall you would expect chesscom to be able to spot and than ban him again if he cheated in these games. Everyone is talking about all these red flags, but there are so many green flags too clearly indicating that Niemann has always been an extremely good player. Did he cheat? I don't know, but certainly right now there isn't that much supporting the theory other than Carlsens gut feeling. Caruana even looked at some of the games with very high accuracy and in most of them found all moves Niemann played relatively natural, another GM I saw analysed 1 of the games where apparently he played like 30 top 1 engine moves in a row and came to the conclusions that a high rated gm would probably found all or close to all of these moves even in a blitz game and Niemann had much more time to find them in a classic game.
the problem is that many people have a biased approach, instead of checking if Hans is cheating or not, they approach it trying to proof that he cheated or trying to disproof it, instead of collecting all the facts and then maybe drawing a conclusion or maybe understanding that the evidence isn't conclusive.
@@LunnarisLP I understand your point, and I don't think there is any possible way that I could use any device and beat Magnus or Hikaru.
It can be true that Hans is an extremely powerful player and also that he is cheating. In fact, I would expect that to be the most likely case.
So his other stats, like bullet chess, would show that.
But the number of times he has had 100% or 90%+ engine top move correlation is too insane to be ignored.
I do not understand those data in any way other than he cheated in many (but not all) of those games.
A 45 move game against one of the top GMs in the world with a 100% engine correlation?
That MUST be cheating.
@@LunnarisLP My view is that we should assume that if a player wants to cheat, they will, and that we will never be able to prove it. Regan's algorithm has been put forward as the leading statistical method for cheat detection, but as described here, it can't detect cheating even in theory.
Niemann is a cheater by his own admission. He has displayed a pattern of cheating over the course of 5 years including in games in paid events, most recently in 2020. We will never know whether he stopped cheating in 2020 or just got better at it. People can decide for themselves whether they want to play against him or invite him to tournaments, but it's a completely reasonable decision to never want to play against him again.
well said.. exactly what I was thinking as I watched.. the analysis seems to be missing a key point if I understood it correctly.. It seems to assume that this guys aggregate rating and most of his matches come from playing honestly, and thus it's analyzing recent matches to see if there's outlier numbers/ROI.. but if he's been cheating his whole career.. (which there's seems a lot of belief in the chess world, as well as his admissions that he did this when he was 16 (only three years ago), then this entire analysis seems null and void
The thing is that Nieman himself admitted that he cheated in online chess, and people think that a criminal mind never changes.
You acting as if cheating on online chess at 12 years of age is the same as being a child rарist 💀
a criminal mind ?!
@@Q-hv2cb Yes, 12 years old, but he also admitted to age 16 as well.
Hans cheated for many years but you all want to belive in fairy tales about hidden talent. Cheater is a cheater.
No evidence my ass. His percentages make Bobby Fischer look like an ibid. You don't have multiple 100% games
There's a 'not seeing the forest for the trees' element to many folks discussion of Regan's model. His analysis is extremely rigorous and well-executed, but it is limited in scope: It can detect cheating very well as long the cheating meets certain expectations. This is inevitable considering that his statistical model requires a very large data set and, at it's core, compares deviation of sample from expected performance with respect to a leading engine's main line. Plainly, it works as long as the cheater cheats in a predictable way over a certain period of time. A wise cheater with an understanding of such analysis could likely go undetected by strategically limiting his "indiscretions." Overall, Regan developed an excellent tool for *helping* to keep Chess fair, but it cannot do on it's own...especially at the highest levels. Unfortunately the other tools in anti-cheating "toolbox" are clearly still in early stages of development.
Precisely. It's a broad level analysis of the full data set basically to detect the equivalent of a statistical rising tide caused by systematic cheating. It won't detect cheating if it is dictated by other contextual or external conditions. For example, if the method is restricted and doesn't have equal opportunity to be employed then you could have a data set where of all games the opportunity to cheat wasn't large enough to create a significant statistical deviation across the whole set.
Considering that Hans has 10 games on chessbase with 100% score and a bunch more over 90% it is puzzling to me that Regan's algorithm didn't catch anything wrong. No other player even has one 100% score and very few have one over 90. This is super fishy.
@@gametime2473 this isnt true btw. Hikaru has a 100% accuracy game and so do several others
@@tristan6066 Ok, I haven't investigated anything I just watched the Yosha channel video. She said the highest correlation score before Hans was 98% and that player got busted for cheating. It seems really odd that Hans would play that many perfect games even if he was cheating. He is basically narc'ing on himself doing that.
@@gametime2473 She compared his individual games with tournaments or sometimes even multiple years for other players which makes no sense. And ignored the main reason Hans was able to play this high accuracy games, it was because he was playing way lower rated players, which he did because his own rating back then was much lower than his actual strength. The fact he already was that strong at the end of 2020 when his fide classic rating started to go up is clearly visible by e.g. his chesscom rating already peaking there for bullet and blitz, in case of bullet chess he was ranked top 5 at some point. Given the amount of games played I find it hard that he was able to keep the rating while cheating and on top of that not getting detected especially given chesscom likely paid extra attention to a guy they had just banned like a year before or sth.
"You wouldn't cheat just one game" Well, for quite a lot of people, beating the world champ would really be up there with "You would only cheat to win the tournament or financial gain". Other examples where your conclusion fall flat is if you basically the 2nd best player of the tournament and you only need the cheating against that one player above you. Last but not least some people has a nemesis, pure emotional hate/dislike of another person.
Hans wasn't 2nd best. Hans cheater multiple times and was caught 2x. you don't get ca7ght 1st try
You can list a lot of hypotheses that someone did something for some reason, but this does not constitute any evidence in favor of Hans cheating.
@@Tarotainment he never said Hans was 2nd best. He said if you cheat against the 2nd best player, you'd only need to cheat once, especially if you can beat everyone else. If you're the third best player for example.
@@idontno0 close but not quite right. He was saying if you are the 2nd best you'd only have to cheat against the #1 player.
Nothing indiciates Hans cheated against Magnus. It was not Nieman playing super great there. Magnus had an Engine correlation of 41% in this game, which for him is just super bad.
The thing is that if you are already a good player, the crucial times you need to cheat is very low. Statistical analysis will only see cheating patterns if the cheater in question cheats a lot. A cheater cheating one or two moves in a critical position in one or two important games in a tournament won't be noticed in such analysis.
By that logic we should suspect everyone, including carlsen
Dude, he exactly adresses this in the video
@@kartikeyatiwari2502 No. By that logic, statistical analysis of checking for cheating is a useless tool. If you want to find a cheater, you'll need to find actual evidence. (ie, find tech and other signalling solutions.) And in absence of this, suspecting anyone is a lost cause.
@@Asrudin "By that logic, statistical analysis of checking for cheating is a useless tool. If you want to find a cheater, you'll need to find actual evidence. "
You actually need to find evidence the case in the video where he suspected someone of cheating. They didn't say "Ken Regan says statistically you are guilty.... so..." They watched him carefully after his statistics were shown to be suspicious. Useless no, smoking gun evidence fuuuck no. Like a Mammogram, it shows indications of an illness, but you need more testing to prove it.
9:05
Thank you for taking the time to do this.
My pleasure!
2:28 Brazilian GM Leitão (seven-times Brazilian champion) here in Brazil says that at this position the best move is cxb5, meaning, c5 is not a "forced" move at all, instead requires a lot of calculation.
What I suspect Mr Regan means by "forced" move is, for example, when I trade Queens you take my queen and I take yours. That's an obvious move, not a "forced" one. What I think Mr Regan does, is to exclude obvious moves from his statistics. These considerations are just irrelevant to whether or not Mr Neimann cheated and only have to do with Mr Regan's method itself.
As to Mr Rousis's method of cheating we can only conclude that he is a "dumb" cheater. A "smart" cheater wouldn't commit the same mistake that Mr Rousis did. He would oscilate his performance the same way Mr Regan showed in his statistics to be "normal". So, the Fair Play Commission didn't confirm that Mr Rousis was cheating due to Mr Regan's "excellent insights" (although they eventually were proven right) but due to the fact that they found a phone in the toilet that Mr. Rousis had had installed.
8:45 Here it is stated that one player's opponents were playing the best possible chess while Mr Niemann's opponents were playing very bad chess. So, I ask myself, if such unpredictable counter-statistical performances can happen in chess, how on Earth can I trust statistics to make final decisions that so deeply affects a player's career?
My conclusion is, a "smart" cheater, as opposed to what was stated in the video, is a player that cheats with "systematic consistency" but at the same time makes sure he closely adheres to Mr Regan's expected statistics. This means that Mr Regan's conclusion that "there's no reason to suspect that Mr Niemann is cheating" should have a caveat: "unless he is a "smart" cheater and closely sticks to my statistical patterns of "normality".
Really nice and relevant points!
The highest rated Brazilian Sunday night TV show "Fantástico" just aired a thorough report on the Niemann saga. Check: th-cam.com/video/iCSy96amvMk/w-d-xo.html
It's Magnus who statistically played multiple SDs below his rating against Hans. Statistically, it looks more like he intentionally threw the game, than that Hans cheated.
Normality is defined as what other chess players at that rating would play... And comparing that vs what the engines says.
Ken has no input on what IS a forced move... GMs and Engines define that by their play alone.
Also, it's not black and white, some moves are critical and hard to see (the ones a "smart cheater" would cheat at), and others are forced... ie, most GMs would agree and the engine also... and there's a whole spectrum in between... Like when 2 moves are about equal, and most GMs find both... Or there's an only move that nobody finds but the engine does... Etc.
That's what Kens method corrects for... Which is exactly what you guys are arguing for... and you are correct in arguing for it... it's a sensible thing to argue for... but you are missunderstanding and understimating Ken's work since that's exactly what his method solves.
Ken's method would find (and does find) this supposedly undetectable "smart cheater" because it can decide which moves were critical to cheat at, and how hard it was to go right and it corrects for it. and over the long run you'd be found.
The fact that GMs think that they would get away with "cheating smartly", just tells you that they aren't that good at stats...
@@azlastor I don't think many GM's believe in undetectable smart cheating ... it's mostly chess fans
Hang on. Havanna isn't the most suspicious tournament even according to the Regan's list. The first is Charlotte's CCCSAFall. Any chance to see that detailed analysis? BTW that's when HN got his last GM norm.
I work with statistics too. This methodologý is good for amateur cheaters. But the main issue is that a smart cheater would need more games to spot the help, he can even use the help sporadically and even lose the games when using help to show some partial brilliancy. It is the common problem of statistical power.
But if you do that, you don't benefit from cheating.
Niemann has a perfect bell curve... Like, as non cheated stats (in the analysed games) as you could hope to get. It's actually funny.
This only looks at the moves so it doesn't look like it even cares whether or not a player wins or losses. Also I think a player making a few 'brilliant' moves but still losing would be suspicious in itself. To be assured to lose they would also have to make sure to make a few poor moves to let the opponent win. When someone in chess is playing well it's unlikely that they're going to make a few brilliant moves, but then also make the kind of blunders necessary to assure they lose. If someone plays a game really well, making some very good moves, they will rarely lose unless they happen to go up against someone who simply plays an even better game. Which is something you can't really plan for.
@@azlastor But isn't his ROI calculated based on his current rating. As his rating improves, the bell curve moves to the right resulting in values that might indicate he is not cheating. Eg If I cheated and moved my rating from 2000 to 2300, the ROI is calculated against my latest rating namely 2300 when in fact I am just a 2000 player)
Okay good addition. Reading this comment it just bothers me that so many people seem to make incorrect assumptions about the model to indicate that he was cheating, I mean statisticallly (lol) many people are making conclusions based on intuition here, and that’s a problem with how most people understand statistics as I’m sure you know. I have a few short questions
- Is the claim that “saying that Hans opponents just played worse in a nonsensical, since if he did engine moves they’d naturally play worse” something you’d agree with, or do you think there’s issues with this portion of the model besides that (to me as a layman this ignored the example of the player that played well but played against other people that played well, therefore lost rating, since the methodology for this part of the model wasn’t delved into)
- does the fact that Hans Neumann knows how knights move prove that he’s cheating?
I meant “reading these comments” and “Neimann”, going to stop commenting so much on this client since I can’t edit, lol.
Ken Regan analysis is a joke for Caruana. Ken Regan wanted to avoid false positives but that gave lots of false negatives...
Any algorithm that produces false positives present more danger than algorithm that produces false negatives.
Huh? Prove em wrong
As I remember Caruana is a professional chess player, not quite an expert in this field 🤦🏻♂️ Fabiano's opinion has absolutely zero validity
Sorry, but vaso is right.
@@smartfck4 but he had a good point: we know when Hans cheated, he admitted it, run the Regan analysis for this period of time and see will he get caught?
What isn’t taken into consideration is the massive deviation between how poorly opponents played vs Hans vs how they played vs other opponents. The caster here simply stated “they didn’t play as well” as if it was just luck. If you’re playing the best moves at the right time, it will force your opponent to be innacruate. That number being so much higher should be a part of the analysis.
For example, I bet in the end game vs magnus during the most famous game, Magnus had a few inaccuracies…whereas he probably had little to no end game inaccuracies vs others. Why is that? Luck? No. It’s bc perfect play causes innacuracies. Think about it
That would be visible in the roi... the roi was normal so he played according to his rating.. so he didn't cheat and was indeed lucky the others played bad against him. Analysis of the game against magnus also shows that magnus didn't play very well.. it is literally just a world champion who is buttsore he lost
yea I typically also lose because I make mistakes. saying just play the best moves against stockfish is just .. dumb.
If you're playing the best moves it does not force your opponent to be inaccurate. This sounds ridiculous.
@@jay31415 if you play the engine you will always make mistakes, I promise
That's an illusion. Stockfish is more likely to exploit your mistakes, but you're making mistakes even when you play a beginner. It doesn't matter who you're playing. On each move, you make a choice. That choice either aligns with a top-engine move or it doesn't.
I'd be interested in analysis of why people tend to play badly against Hans. I imagine it probably has something to do with his playing style and maybe even psyching out his opponents, but I haven't seen anyone go in depth about that
In my opinion is the same motive why his results are inconsistent. He is aggressive and tries to push the game out of the opponents preparation into unknown and complicated territory. That certainly makes the opponents play worse, but it may work against him if the opponent calculates better than him in a specific game.
Well there are players who are known to have these kinds of qualities such as Tal or Shirov, but a grandmaster friend (and former US champion) told me that his opinion of Hans's style was that it was raw and unpolished, and that was probably irritating to more established players.
Are you new to the chess scene? Hans has been losing badly to the top players on all time controls until these last 2 tournaments. It really is an overnight improvement in performance
@@albertopulido4267 don't be an idiot. Magnus has shown why he is on top for over 10 years even since before regular engine use was a thing. He was the youngest gm at the time and drew against former world chess champion kasparov at like 10 years old. Everyone could see even at a young age that magnus was going to be a top player. You make yourself look ridiculous comparing him to a known cheater who up until 5 minutes ago was known more for his online personality than his chess ability
‘Playing style’ & ‘psych’ lol. Talk about a stretch / reach. Cause these super GMs who have seen almost everything crumble upon the slightest deviation or are totally unable to analyze some different situation.
‘Psyching out’. Yea, cause a guy who has fought through the pressure of 5 world championships gets all out of sorts cause of the supreme death stare of this amazeballs 2600 elo 19yr old who has won everything under the sun (o wait).
Regarding the ROI argument (partially addressed at the end), has it been considered that perhaps Niemann has so many performances “below his rating” on purpose? This could be a designed attempt to pass these statistical tests undetected. For example, most of the time Niemann could strategically underperform a bit in “unimportant” events, and when he gets an important event in sight, he could cheat in a cluster of events to qualify and then win the one he was aiming for. I think I saw a different statistical analysis that did show clusters of outstanding performance in time for Niemann, but I don’t remember well.
One of the arguments Niemann has made for his recent rise is that he has been playing many games and tournaments, but this could also be part of a strategy to beat the long-term statistical analysis. More games = more chances to compensate for outliers.
Thoughts?
Exactly!
what you're describing is doping or micro doping
Xe_None, my thoughts exactly.
The algorithm should be if innacuracies are played at the beggining when the position gets more complex if better decisions are made, it should raise suspicion at the very least
Apart from all the drama, this is super interesting! I wonder if ROIs show different playing style as well (for example a player who plays more non-sound moves but invites the opponents to make mistakes and thus gains a relatively high rating for their accuracy levels vs. someone who plays very solid, a lot of best moves, but draws a lot).
You mean like Fischer playing Spassky in game 13 of the 1972 World Championship Match.
Yes there are other factors as well, such as Hans saying that he happened to be prepared for Magnus's line by looking at a game between Magnus and So. But that game in question was a blitz game, and the idea that Hans would try and memorize the line 14 moves deep from a blitz game before a tournament is suspect (if this were a match against Magnus it would not be suspect). Then, he kind of fumbled some onboard analysis between his game with Firouzja, I believe, and he refused to go into analysis. Right after the win against Magnus, he also refused to discuss the game, saying "chess speaks for itself". You'd think Hans would be excited to speak about his great victory. Most players love to go into variations and discuss the game they won. I saw Ivanchuk crank out variations in his head when being interviewed over a couple games. Yet Hans say, "chess speaks for itself". Apparently not, Hans.
Suspicious, yes; proof, no
@@leftjedi
It's not just a little suspicious. It's like coming home to you're wife buttoning up her clothes looking extremely nervous with a unknown pickup truck parked outside and a history of cheating.
Being a zoomer and having a different personality to chess greats 20 years older than you isn’t exactly cause for suspicion. Got to remember chess speaks for itself line was before anyone talked about cheating, and he was probably just having a laugh and being an edgy teenager.
@@davikarlsigursson5657 this would make sense if it WAS LIKE that, but it was NOT. two different scenarios. This is called the straw man attack, you build a totally different scenario and argue it, and of course it is waaaaay worse, or at least very subjective, but NOT same. He surprised everyone, including Magnus, a notorious sore loser, by beating his ass and making him look very bad, and so Magnus also no builds a strawman attack to remove his embarrassment, and now instead of the world talking about how BAD Magnus got beat by a lower player, we are now talking about the STRAWMAN, how Hans is a cheater. Hans won the battle but Magnus is a master strategist and he is still fighting the battle. In the end I believe we are seeing the END of Magnus reign, as his worse fear is here, a player, a young player who has his number and prepared, created himself as the person who will dethrone the king. All hail the new king, Hans Nieman.
@@normanlove222 he was a nobody who cheated and didn't have the best rating but yet he sacked the best player to ever play. You don't just come out of nowhere and do that especially when you have history of streaming and you getting destroyed. Including the beach games against Magnus which Magnus destroyed him. There's pictures to prove it and a GM that watched it. Enough is enough.
Assuming it's much more difficult to cheat over the board, it would be good to see a comparison between those stats with the online game stats.
The analysis from the previous video showed that his worst tournaments where online
@@answeris4217 that's interesting.
@@answeris4217 because he isn’t cheating
@@tajneeley look I can't say he is or isn't. That data sure look that way
I mean they should step up the anti cheat methods after this drama, i dont see how its difficult to make it impossible to cheat when playing face to face, search the players, scan them whatever, its rediculous that there is a possibility to cheat in chess.
Once you know about ROI All you need to do to crack it is to tactically throw some games on purpose. As evident in Norman's performance
Exactly! and:
"
10:16
I absolutely disagree! and think you are biased.
It would absolutely give you the result you want:
More rating, more money and fame and without the risk of being caught.
It's the move of the "very smart cheater".
You don't have to win the tournament to improve your financial state.
As long as you have a better rating than you could have had - then you are improved.
"
I'd like to know more about how Regan's method has been previously used. Rousis' case is a prime example of it's success, the analysis flagged him as suspicious which was later confirmed by physical proof, but how often does a player shows suspicious results and no cheating can be proven? And how often does a player cheat and isn't detected (yeah, I know, this one is pretty hard to get data on). In a perfect world you'd get 100 GMs to play a mock-tournament and have 50 of them cheat so you can validate the algorithm against a known reality.
I would like for that event to happen in public, and for the spectators to be part of the show: they win something if they correctly guess who the cheaters are. Probably a bad idea, though.
The problem is that regans method is and has been fully known and disclosed for over 15 years. All a cheater has to do is a build a system around the known parameters of his anti-cheating method.
Regans statistics are very good, it’s just that current method is outdated
Disclosing the details of a cheating algorithm is "outdated".
Yosha came to the opposite conclution, and said there was 1 one to 77600 for his good results in 6 tournaments on a row, and he had at least 5 games with 100% of his moves equal to the computer.
Boom!
Her analysis is completely flawed.
@@cartifan399 Where did she go wrong?
@@bjorn2fly everything was wrong pretty much. If you do the math correctly on the ROIs she was showing it comes to about 23 in 10000 which ~1 in 500 chance of him having those tournaments (very possible if he recently improved due to a covid break). The 100% correlation she was showing was with any of top engine moves (could be from any engine, so each move from a different engine) which many grandmasters including Hikaru, magnus, etc have had. I am honestly very sus of Hans, but the analysis was useless
@@rishabmeh32 And her analysis used Ken Regan's analysis then completely ignored anything that led him to his conclusion.
Good stuff, Albert. Makes things a lot more understandable.
No it's not.
It's a biased an illogical with wrong conclusion and way of deduction,
With clear logical mistakes in the analysis
Nice follow up! This should work VERY well, most effectively against the “smart cheater”; albeit over hundreds of games necessary for data. Even just one critical move every few games would still show up as a significant deviation in ROI over time.
EXACTLY, congratulations for understanding that!
Also, IMPORTANT! HANS data is an almost PERFECT bell curve..
It's not a case of "this data is funny but not skewed enough to call him a cheater..." NO... It's as non cheated a dataset as you could get, that's why Ken is so confident that over the games analysed Hans did not cheat.
This is an explanation that chess needs. No biased arguments, no hasty judgments, just clear, concise and transparent information. Congratulations. Faced with the scandalous and the fanatics on one side and the other, what must be opposed is the concrete data.
No it's not.
It's a biased an illogical with wrong conclusion and way of deduction,
With clear logical mistakes in the analysis
@@אלעדר if you say so, I don't know why someone calls themselves Albert or God should think something, it would be enough to consult you.
First if all I'd loved to see a different sorting being applied to the ROI statistics as it was unclear if there is any kind of time-based trend recognisable.
Secondly, for this very specific case, wouldn't it be a valid approach to take all of his opponents, too, and undergo the same analysis in order to have a clear indication if their ROI was just completely off that day and how this compares to their former performances.
yeah, exactly: specifically the second part should be done. ROI would be useful only if someone was constantly cheating
Ok, at a certain point, would we be more impressed with Hans if he figured out how to beat the statistical cheating detection algorithim, and cheat OTB without detection time after time than if he had just beaten Magnus of his own volition?
It wouldnt be that impressive tho, honestly if he gets caught by reviewing his previous games he just messed up and overcheated. A GM should be able to cheat a couple of moves per game and not get caught. In his case, some of his entire games have been very precise engine lines. It would be impossible to catch a "smart" cheater that only uses the engine moves sporadically. If we do catch cheaters without actually catching the cheating act (OTB device or an engine running online), well, they were pretty damn stupid in the first place, that's the scary part about the future of chess, people just might boost their performances occasionally and there will be no way to know
Garam - the main point of the video is finding people who cheat on only a few moves. All the people who detect high level cheating know about this and factor it in to their analyses.
hi, i want to share 3 points i was thinking about
1- beside the forcing moves that shouldnt be included in the ROI there is also
- moves due to intuition
- moves due to prep (not only opening moves) some positions are studied using an engine to the 24th move. if the opponent fell into your prep and you're blitzing moves like hikaru in the condidate tournament then he will tend to perform worse while you're performing better
- psychological moves .if you know your opponent is
- under time pressure
- under pressure of performing well in a tournament to get access to condidates for exemple or the next tournament ( in the last condidate tournament caruana couldl have drew certain games but decided to go for risky plays because he was behind in points)
all those 3 type of moves could influence your opponent roi and yours and we hear of them too many times by top players, so they should be taken into consideration
2- cheating isnt just about finding the best moves. just knowing the eval of the position can tell you if the sacrifice made by your opponent is worth it or not. for exemple alierieza avoided a loss against hans by trusting the sacrifice hans made ( obviously thinking hans is using stockfish so if he sacrificed a piece that means there is gains in the position)
3- if the player is just improving, wouldnt that affect the roi
I guess for your third point, if the player is getting better, the roi is calculated for the new rating he has. So it would still be accurate.
Now please use these same methods and show the period of time when Hans himself admitted to cheating. Otherwise you're not comparing apples to apples
Very good. Yea this would be great to see
When he was 12 and 16 he cheated online and was caught by their algorithm. I dont see how that would be at all similar to OTB?
@@statictech7 The tool they are using to analyze his games doesn't care whether it is online or OTB. So comparing the output from his cheating period to the current period would yield some useful information. Much more useful than comparing it with some other known cheater as they did in the video imho. And depending what the results of the analysis are, it could either help his case or hinder it. Either way it would be interesting to see the result
@@fufu77 But there is no reason, that Hans should cheat the same way now...that is, if he's still cheating.
@@fnjfrancis There is a reason to run the test. It's called a control test. We know Han's cheated then, so to test this algorithm, it should also yield a "positive" test.
Many engine "top line" moves are unfathomable to even GMs. The analysis goes to 20 moves or more depth, and is obscured even more by the neural net. So even if a player sees a top-line engine move he may still not understand the engine "plan" for that move and so playing it would be counterproductive.
If you only play one, perhaps, but if you play 5 or 6 moves (even alternatevily) in a match after classic openings, the thing changes quiete a lot. Besides the computer thinks again over your mistakes as its new position. Just play with the help of the computer with a friend, having yourself 5 options to receive help and he not 1, and you will see how you beat him quite more easily at the end of the day. it will be not counterproductive I say. Besides, when you know that is really a good move and you are a GMs is still more profitable, so you can understand it easier and quicker if it is humanly possible.
@@Buymecheap yes. I think your answer is more accurate.
@@Buymecheap
If your answer is to get cheating help for many moves then you are negating the starting premise that states that all that is required is perhaps two or three moves.
Unless he/she is fed the following moves also! ;)
@@howard5992 No, I Just say that normally any help anytime is fine. And whem more moves are indicated to you by the computer so the better.
Based on what you said at the end of this video, am I following you correctly that Regan's method won't catch someone who only cheats in very limited circumstances? Someone who is mostly playing honestly but cheats on a specific move when they perceive a particularly high payoff?
Exactly!
and:
"
10:16
I absolutely disagree! and think you are biased.
It would absolutely give you the result you want:
More rating, more money and fame and without the risk of being caught.
It's the move of the "very smart cheater".
You don't have to win the tournament to improve your financial state.
As long as you have a better rating than you could have had - then you are improved.
"
Kudos to you for pointing out what so many of the other videos have been missing: you always have to look at the quality of the opponent's play as well.
Yes, if Nieman has a magical psychological ability to make his opponents under-perform (maybe because they were under-estimated him. "oh today I play with this kid with the funny curly hair?") , then it is something that can help his results.
Perhaps the suspicion of his cheating caused his opponents to play worse. In the game in which Magnus accused him, Magnus played well below his usual standard. He was clearly distracted by the cheating issue.
Your conclusion "Statistically there is good reason to believe that he is playing honestly", although carefully worded, is not correct.
The analyses done by Ken Regan, and he would undoubtedly agree, only lead to the conclusion that cheating is not proven, not statistically, but that only according to the particular statistical methods used. That is all. This conclusion is entirely dependent on the particular statistical analyses used.
There could indeed be other statistical analyses leading to the opposite conclusion. Nobody can exclude this, and certainly not a serious statistician. To conclude as you do, one would have to establish that this is impossible...
On the other hand, the same kind of reasoning leading to your conclusion would lead to the conclusion that before 2012 there was "good reason to believe" that the Higgs boson did not exist! No, it just hadn't been detected. HN could be a kind of Higgs boson... :-)
Finally, knowing the statistical methods used by Ken Regan, it is possible to cheat without being detected by these analyses: it is enough to couple the chess program used with a program ensuring that the statistical distributions will be consistent with those expected. This argument alone is also sufficient to refute the possibility of reaching your conclusion.
That's just not an accurate way of checking if someone is cheating. I actually thought that Ken Regan would be going straight to the source. Cheating in modern times is almost always about using a chess engine to give you your next best move. As such I would expect that an anti cheating algorithm would connect directly with the top chess engines out there (ie Stockfish and Alpha zero) and then correlate every move with the engines' best moves (while also ignoring openings and forced moves) and then yield a correlation percentage based on how many moves match these engine's top moves. That's what made sense to me.
Instead, Ken Regan is using an engine to tell him if a player is cheating based on his assigned rating? Hans has been cheating since he was 12, which means that you simply have no idea if his rating should be trusted in the first place.
As a thought experiment, if you were to lower Hans ranting, by deducting 200 points his assigned rating at any given point in time, and run the analysis again, how many of his underperforming games end up above 50 ROI?
Your straightforward way of directly comparing with engine moves would probably have a lot of super GMs branded as cheaters, based on where you put your cutoff value. And trash players like me could cheat a whole lot without being found out.
The problem of the long-term cheater is pretty interesting. But first, if Niemann, like you put it, "has been cheating since he was 12", (semantically not wrong, but pragmatically implying that he's been cheating pretty consistently all the time) how do you explain that his extreme rise in elo is a lot more recent than that?
Really good data but I'm stuck on the fact that all the players that played Hans all played below their norm. How is that possible? Is that normalcy?
You said that with Regan's method, only a very limited number of moves per game are relevant for the analysis and that you you could cheat on some(none relevant) moves without being detected. So that makes me wonder, if it is possible to figure out which kind of moves get ignored by the analysis and systematically cheat on those moves. Because as other people's analysis shows, that there are indeed factors in which Niemann clearly and consistently deviates from the majority of other (most notably stronger) players.
That alone makes it completely worthless for detecting any high-level cheater.
yeah but why would you want to cheat in a position where there are like 8 moves giving you e.g. a engine evaluation starting from +0.4 down to +0.2. If you want to cheat your way up cheating in the important moments is what you need.
Yahhh, there were so many holes in the logic of regan’s selected data points that I’m no even going to waste time itemizing the issues.
The moves that are ignored are typically moves in which cheating would not matter e.g. forced moves, opening theory.
The only possible one is when there are multiple winning variations, which, as said in the video, it is not uncommon for a highly rated GM to find one.
The moves that are ignored are ignored for a reason. If someone makes moves that make a significant difference in the game, those moves aren't going to be ignored anymore.
I watched Ken Reagan's analysis. I couldn't make head or tails of it. Thank you so much for giving a quick simple take on it so I could understand.
I am more confused than ever. There was a recent analysis of Hans' games and it showed that Hans had 100% accuracy in TEN games on chessbase in OTB games. No other player in the database even has one. One player got a 98% and was caught cheating. How does Regan arrive at the opposite conclusion?
@cDb I got it from the Yosha channel. She did a breakdown of Hans' suspicious games.
@cDb lol, yeah. As a GM I would figure that he would be aware enough not to send up red flags left and right. Who knows what this "correlation score" is and other people have said they ran a correlation for those games and they weren't 100%. Statistics are so easily manipulated.
That's wrong, many players have 100% games. The comparisons made in Yoshas video are over much longer time periods or at least whole tournaments, comparing them to a single game is unfair. The comparison in itself is unfair too, because Hans was playing people rated much lower than his skill level (if we assume he is legit) so achieving a high accuracy is much easier compared to when super gms are playing against other super gms, which is the reason why for example Hans never had and will never have a 100% or likely even a 95 or 90%+ game against any of those super gms. But against 2200-2400 rated players it actually isn't that weird and GMs like Caruana looked at the games by themselfs and found most of the moves to be rather human and natural.
Because Regan is a hack and his method has glaring problems
@@LunnarisLP many players we are talking about here - GMs - are lucky to even have a 90+ in their entires careers. “many players have 100% games…” LMAOOOOOO 🤣🤣🤣
FYI, Hans is a known cheater, just not OtB. I don't agree with the argument that the motive for cheating is to get a high placing or to win for the financial incentive. Hans had not until recently been invited to any major tournaments. All he wants is to continue to make these large tournaments. He is after the fame, and as a content creator, the financial gain that comes from it. All he needs is the occassional win against top players. Good luck catching that with statistical analysis. In his game against Magnus in the Sinquefield Cup, after the opening, Hans made all top 3 moves (according to stockfish) and on three occassions made the only winning move. That paired with his refusal to explain lines and variations in the post match interview is incredibly sus. With a history of cheating, his quick rise in rating, and considering his age he should be scrutinized.
the big question though is how the hell can he cheat in OTB? OTB anti-cheating regulations have been extra strict since the Toiletgate scandal between Kramnik and Topalov... some players have actually complained about that... Hans must be using CIA-type technology to cheat in order to evade those anti-cheating checks
Ha ha even M Tal does not know all the lines when he made sacrifices and he admitted that. Sometimes you can't explained all the lines when you made a move, just relying on intuition and feeling for the position. Trying to dig all the lines and variations during the game will lead you to time trouble in most cases. And what would you expect to Hans when it comes to media interviews ... he is very young and less experience in responding to top level media interviews that's why he can't explained many things clearly and made some erratic statements.
@@excalibur92 according to your logic, even tal doesn't play 100% accuracy. This hans guy won with 100% accuracy more times than Tal, Bobby, Magnus, Morphey combined. Let that sink in, trust me, he's not that great.
@@excalibur92 Tal and his opponents didn't have access to strong chess engines that are available today. His playstyle was based on previous chess knowledge and intuition. And there is also the intimidation factor. His opponents might have just assumed that they are worse after he played some sort of a sacrifice and they might have lost their concentration.
When it comes to Hans's interviews, not only did he not know how to explain his moves but he also said that his position was clearly better when it wasn't clear at all. Even Hikaru Nakamura admitted that during his livestream. You have the right to be suspicious of somebody who has cheated in the past, was banned from chesscom and whose coach is also a notorious cheater.
@@maksymiliank5135 Many top GMs today when they played a move and ask to explain it after the game ... you can always hear from them the Russian expression "I don't know" when it comes to evaluating and analysing their games. Then maybe they can also be accused of cheating?
This is really really well done. Pleasantly delivered, calm, sober, friendly, straightforward, knowledgeable. It's a pity we're getting this only by now. Congratulations.
No it's not.
It's a biased an illogical with wrong conclusion and way of deduction,
With clear logical mistakes in the analysis
@@אלעדר explain what's wrong with his conclusion, deduction and logic. In addition explain why or how he's biased.
@@zaimnaqvi8893
"
10:16
I absolutely disagree! and think you are biased.
It would absolutely give you the result you want:
More rating, more money and fame and without the risk of being caught.
It's the move of the "not a complete idiot cheater".
You don't have to win the tournament to improve your financial state.
As long as you have a better rating than you could have had - then you are improved.
"
@@אלעדר This is just idle talk with zero substance.
@@higamato3811 What about the argument, that he could have cheated only sometimes, don't you agree with?
This way he wouldn't get caught by this statistic analysis and also improve his position
Can i know if Niemanns low ROI below 50 are on the board with live broadcasting or without ?
Id love to see an analysis of critical moves.
I think measuring cheating with all moves included (except forcing moves) is still too broad of a scope.
A smart cheater would only cheat at pivotal moments. (Such as moves that help you survive an attack or brilliant moves that will win you pieces)
I'm not sure how those could be measured but it would require much more rigorous analysis of each game.
That point is addressed in the video..
You don't understand what the video explained, your point is explained clearly in the video.
@@excalibur92 yeah and then in 10:10 he admits a smart cheater wouldn't cheat in every game and as a result would never get caught. Regan concludes Niemann didn't cheat, despite Niemann having games 50-60 on half that list like Rausis, the ones where he didn't cheat are used as....evidence he never cheated. The assumption that a cheater wouldn't benefit by cheating only like once in a while in the video is so wrong just by this very controversy and how Niemann's rating jumped to 2700 after one victory over Carlsen
@@jacobpeters5458 I’m not sure how those two things are connected (spoiler: they’re not).
It’s a bell curve. If you didn’t have ~half your observations above 50, then it wouldn’t be normal.
@@jacobpeters5458 No, the video explained his opponents frequently played worse than usual, but only against him. I would wonder if there's some correlation between cheating rumors and his opponents beginning to play poorly versus him, but you would need a source in GM circles who remembers when the rumors first began circulating.
Also, one could just analyze the games and see, but then you are doubling up on work Regan has already done.
What does Ken Regan algorithm say about the games Hans played earlier, during the period he admitted cheating? It's obvious that he cheated more than those two times he was caught. Algorithm should be ringing alarm bells.
So according to many GMs, when you're playing against someone who's suspected of cheating, your performance may drop because you can't focus that well. I believe this is exactly how Hans cheats - he spreads rumours about him being a cheater, e.g. during the infamous interview, in order to stress his opponents and sometimes even get free wins, as he did with Magnus. It's ingenious, really.
This is the flimsiest thing I have ever seen on this cheating topic. By your metric, I can say Magnus’ title as the best in the world is what makes other people play worse. The pressure, and the stress of having to perform.
In fact my example is far more likely than yours.
@@somethingcool9979 Yes, this is correct! I'm glad you've noticed the joking tone of my comment and provided an even more hilarious example.
@@anatolydyatlov963 I was 50/50 on whether to is was sarcasm or not, because this wasn’t too far from what a lot of the chess community would actually speculate lol. So I decided to bite, the last three words would have confirmed the joke but I was at work when I saw the comment and didn’t read it entirely. RIP.
@@somethingcool9979 Hahah no worries ;D Conveying sarcasm through text has always been a challenge.
Then he can sue them for the accusations. Brilliant!
Centipawn model is far superior to this when it comes to analyzing expected rating vs. actual rating and there it is clear cut that Hans cheated. Why come up with a much more complex model when the more basic stuff is clearly superior? As a Mathematician myself I can answer this: it is because the simple stuff can be covered in one published paper and that's it, there is no milking that cow. Coming up with a more complex model even if it is worse, will allow you to milk that cow over and over by further refining it which will subsequently allow you to publish more peer reviewed papers. If anyone doesn't know: the quality of a professor at a university is usually measured in the number of papers per year, not in the quality of said papers, this gives a huge incentivo for people to publish trash research just to increase their productivity instead of publishing great stuff. Also as a tip, if your research is good, your academic credentials don't matter (Einstein was a post office boy), so starting the discussion with "this guy is a PHD of X, from the University of Y" this is a huge red flag of someone trying to appeal to authority BEFORE presenting their argument which could indicate dishonesty.
Ridiculous. If cheater knows algorithm that is used, he can easily make
the strategy. There is also random cheating approach. So, total BS. Not saying that Niemann cheated in the over the board tournaments. But his history of cheating online would always haunt him. Him admitting cheating twice and only when he was called speaks volume too.
Easily make the strategy? No, he couldn't. Must be nice to have such a big ego that you think you know better than the foremost expert on cheating. Admitting he cheated speaks volumes about the fact that he has changed his ways. The fact that he cheated online when he was literally a minor was well known in the chess world long before he spoke about it publicly. No one would volunteer that publicly if it wasn't necessary.
Where is the analysis of the game versus Magnus? Seems like it would be worth including.
Thanks for your unbiased analysis of the analysis. I enjoyed your presentation and look forward to more videos!
I'm not a high rated chess player, but it seems to me that when I watch Hans' games he always seem to throw in a few curveballs in most of his games. He doesn't play the straight forward theoretical moves all game and I think this throws off his opponents. This is not to say he's a better player or faster at calculating lines but rather he's confusing or tricking the opponent. These moves he plays are not the optimal moves so they can give the opponent advantage if they calculate a reply properly. At first I thought he was probably cheating because of the way he so easily beat Magnus, but the more I see from him the more I'm convinced he probably wasn't cheating. He's just figured out a different way to play and right now he's got the advantage before other figure it out and begin to counter his play. I really hope it comes out that he isn't cheating because it'll make chess even more interesting. Really want to see how long he can keep up these great performances before or if anyone can finally figure him out.
if i will play against an engine it is obvious that my accuracy will drop, u cant say that his opponents played badly against him in order to prove his innocence.
Exactly, and:
"
10:16
I absolutely disagree! and think you are biased.
It would absolutely give you the result you want:
More rating, more money and fame and without the risk of being caught.
It's the move of the "very smart cheater".
You don't have to win the tournament to improve your financial state.
As long as you have a better rating than you could have had - then you are improved.
"
I’m glad this video was made to explain Reagan’s results. I am not personally satisfied with the rigor of the analysis. Perhaps there are more metrics that are not being discussed, but there should be more than one method of determining whether or not someone is cheating. Caruana suggested Reagan apply his methods to the period of time that Hans admitted he was cheating. Caruana suspects he would not detect anything, hence the need for those with the most to lose to demand better standards. Modern times call for more creative solutions. Otherwise chess may fade into the black.
"100% of his moves were one of the top three engine moves"
Are you saying no other GMs have that?
@@LegendLength This message was deleted for misleading info.
@@JosBroder Dude i thought you were talking about a proper analysis. She has been debunked hard.
I'm not trying to blow you off or exit the argument. It's just a really sad hit piece by her and it's been discussed a lot at reddit etc..
@@LegendLength fair enough. I’ll look into that. If you’re correct I’ll delete my original comment so as not to spread misinformation.
@@JosBroder Good on you, glad people are keeping open minded here.
The main 2 issues were: The engines vary depending on the player. Hans has hundreds of engines used for his moves, whereas Carlsen and everyone else only has a few. That is because of the way that system works (explained in more detail elsewhere).
2nd issue is she multiplied raw odds in a very amateur way. For example if he played 3 great moves in a row she would say the chance is 0.1 * 0.1 * 0.1 = 0.001% . But you can't do it that way.
I was interested in Regan method since Patrycja Waszczuk case, thanks for explaining. Regan missed important detail in my opinion, opponents, human play poorly when thrown against a computer forced variations, this could be the case of Hans opponents 0,17 accuracy...
Excellent video. Have just arrived at this video after getting back in to chess at christmas, and I'm glad I did. Makes me see this Hans Niemann fellow in a much better light than most of the videos suggest. He could be very hard done by. In either case, the best way to prove his innocence is to realize that there are a lot of people on his side, perk up, and keep playing well
This was a very good and much needed video. When ken was explaining I had no idea what on earth he was talking about. But you explained it perfectly for me to understand.
All that’s left is to see what magnus says in the coming days
Looks like the cat's out the bag.
Magnus already said more than enough with his behaviour. Congratulations Hans with your results!
No it's not.
It's a biased an illogical with wrong conclusion and way of deduction,
With clear logical mistakes in the analysis
In my opinion, the table at 6:15 is the key. If all Niemann tournaments sort by date, it is possible to look, if his ROI is strong correlated / dependent with date. If yes, it means, that Hans starts significantly cheating in specific moment. But if there is only weak correlation (or none), then it looks to be OK. I dont say, if Hans is cheater or not. I only suggest the statistic experiment, which could better explain / uncover this unclear situation.
The ROI analysis is pretty well useless unless wild assumptions are made regarding the nature of the cheating. Realistically, you would ONLY see ROI discrepancies like Rausis's if the cheating was consistent. Its pretty clear that pretty much all players agree that Neimann doesnt cheat every match, or maybe even most matches - he frequently performs quite poorly at tournaments.
If you cheat very infrequently, then it would actually be MORE likely for the ROI to show you as playing "under your expected rating" if you are playing a typical match without assistance.
At the end of the day, its not really strong evidence either way. Theres just way too many unknowns to draw any real conclusions. The only thing this analysis shows is that Hans didnt cheat the exact same way as Rausis.
Very good stuff. But personally, based on what I saw, Magnus had prepared an opening that was unique in his mind. But Hans said in post interview that he had been preparing for that particular opening. Wondering if he had a mole in the Magnus camp. But would that be cheating or just 'research'?
I totally agree with your point, looks like something went wrongly in his team, next week we will get the truth
In any case Hans lied in the interview to deflect the question, so he should not be trusted.
...or hacking...or a setup?
In the next days interview, Hans clarified he studied a transposition, which did end in the same position. So not the same opening, but the same position reached from a different opening.
Wouldn't consider that cheating, its been done many times in the past. Only has limited usefulness though.
Is the ROI for the last 25 games above 50? or am I misreading?
Great video - glad to get an understanding of this statistical analysis done by Regan. Thanks.
No it's not.
It's a biased an illogical with wrong conclusion and way of deduction,
With clear logical mistakes in the analysis
Very good conclusion summarising what really happened here. And the details to back it up.
This statistical analysis only exposes certain types of cheaters. And it's very easy for a wise cheater to circumvent.
THIS METHOD CANNOT PROVE SOMEONE ISN'T CHEATING. A lot of people are missing that point and misleadingly summarising the findings.
I think trying to give reason for cheating is flawed. Cheating isn't rational, especially if he cheated specifically to try and impress Magnus for one game. Sometimes cheating is used for consistency or monetary gain, but as you said it would be much easier to spot if they used cheats that way. I'm sure a cheater would try and look deep into cheat detection to try and side step it.
Not saying he did or didn't cheat, but I don't think we have enough to get an answer.
If there is no reason nor any objective the cheating results are mostly irrelevant and can be ignored. If a person cheats in a single games on a whim it is certainly wrong, but it does not impact himself or anybody else in a meaningful way and it can be ignored.
The cheating that must be caught to allow the game to keep fair to everybody is consistent cheating that alters rankings and important events results. This won't be whimsical cheating.
Cheating isnt always rational*
@@Flawlesslaughter And when it is irrational enough to not alter overall results it is mostly irrelevant.
@@conelord1984 interesting point, but you could argue he did it for a single game to get attention for example beating Magnus. But cheating is relevant to who he us unfairly beating even if it doesn't effect his overall rating!
@@Flawlesslaughter If he did it in a single game to get attention on this game he got a lot of unwanted attention. Overall it is not very important though except for the bad publicity for chess and the blame for that was mostly Magnus and not his even if he had cheated (which I doubt). It is easy to discard this game in specific for other reasons though. It was not Hans who played to well, Magnus played horribly. He was in a bad day and that is it.
Additionally Hans is not being suspected of cheating in that game only though, especially by Magnus. Magnus was already hostile to him before this game ansd made hints about his quick ELO climbing.
You've removed the first conclusion, which was that Niemann is progressing in an unusual way, therefore there is a reason for it.
No reason has been discovered why Niemann is moving upwards oddly - it may not be cheating - but that is what to search for; not just "evidence of cheating."
This shows Niemann’s opponents are more intimidated by him than other players.
That’s just not true… Hikaru destroyed him last week effortlessly, and made the difference between a high Grandmaster and junior Grandmaster quite evident.
The only reason high Grandmasters would be afraid of him is if he was cheating, which seems to be the case, at least historically.
With Hans self-admittedly cheating previously along with his coach, can you
blame top tier masters of the game being ultra-suspicious of Neimann and his inconsistent play?
8:12 - a lot of people have already pointed it out that opponents just suspect Niemann of cheating and therefore perform bad, it's not Niemann playing like Stockfish 15
Have you run it on Han's games between the age of 12 and 16 as Caruana suggested? 😂
He won't do that because Ken Regan's method of analysis is kind of a joke. There are players who have been caught red-handed cheating and his analysis doesn't show up as them cheating
As GM Fabiano suggested, Dr. Ken Regan should analyze the games Hans played in which he admitted he cheated. His prediction was that Regan's method would not catch the cheating that was admitted to take place. Frankly, a smart cheater would know not to cheat in every game, on every move. Instead, subtle cheating could be used as both a confidence boost as a safety net. This is why OTB matches should be played with a great deal more scrutiny and security. It's possible that subtle forms of cheating could happen.
It should also be noted that Hans has dozens of games with close to "perfect" according to computer analysis, far above the level of perfection of World Champion players. This is not proof, but considering that Hans was also a late-bloomer, who instead of becoming a GM at 15, he was still cheating at 16 to boost his Chess streaming career.
@@timefororbit Well, two points then. The first is that Dr. Regan did analyze the games Hans admitted he cheated, and Hans came up as having cheated. This is in the report recently released by Chess.com. The second is that the so-called perfect scores are the result of manipulation using Let's Check, except one can maneuver the results up or down accordingly. I have no real desire to do a video on such, but I can assure you I have tactical slugfests by Kasparov, to name but one, with 96% correlation and the like. The thing to remember is that Let's Check is not a single engine. You need only match any of three per move to be counted a match, and you can then supplement them with an engine of your choice.
I don't care about all that science numbers or math. Magnus is saying that Hans is a cheater. That's all I need. Magnus never did this before, so that's it.
And Magnus can never be wrong?
And what about when Magnus himself cheats - Here is the full stream. The events in question are at 1:44:00. Lots of talk of cheating and banning.
th-cam.com/video/CRdrf1Ny3x8/w-d-xo.html
@@mr.safensound4238 Magnus is ChessJesus.
@@Prometheus4096 I really hope that was sarcastic. Otherwise you need to get help.
@graham banks lol
Ken Regan's analysis reveals something very interesting: it's like Hans can nfluence the (playing) behaviour of his opponents. Maybe Magnus felt this influence very clearly, since his chess-thinking is co clear. His coach could not comment on such a 'vague' thing like that.And Magnus cannot 'prevent' something like that either so he chooses nog to play him.
If true this leaves us guessing about how Hans could do that. I cannot resit the thought of 'black magic'.In all sorts of cultures you can go to a 'dukun' (kind of sorcerer) and buy a 'spell' that confuses the thoughts of your 'enemy'. Very hard to trace and to prove.
Good explanation and I think Kens analysis should continue to be used however it’s definitely not sufficient a lot of gaps and the forcing move thing is kind of bs, shows lack of understanding. It assumes its easy to find the forcing moves.
Its part of the definition that forcing moves are easy to find.
Did you look up how the model defines forcing moves or did you nitpick on instinct just so you can confirm your current opinion and move on? Do you think the model defines forcing moves as obscure engine moves that Super GMs would fail to find that make them slightly less losing or equal with their opponent?
@@sams7068 Actually I did look it up but I haven't seen a paper that details the algorithm, which I suspect is not publicly available anyway, however, I would love to read it as a mathematician. However, using the example of c5 it is a fair assumption that the forcing moves assumes a big deviation between top lines and other options. I don't see how "easy to find" can be defined in mathematical terms as the prior reply suggested thus it must have mathematical basis to it such as big drops in evaluation with one/two options available. In the position shown cxb5 and c5 are the only two moves that don't loose on the spot for white. However, if memory serves me right c5 was played rather quickly, when cxb5 is intuitively the more obvious move. b4/b3 with Bxe1 is a big threat in the position that cxb5 actually avoids. Thus, from a chess perspective and not a statistical perspective its obvious that considering a move such as c5 as forcing is somewhat weird. This could indicate that in an incredibly complicated tactical positions such as this one the one or two options that are not loosing on the spot will always be considered forced moves, when we know for a fact humans are bad in complicated positions and shine in positional chess. I also understand the idea of forcing moves in Ken's model as it is an attempt to avoid false positives, but the result could be false negatives. I also have no bias of opinion, my opinion was formed on the video, as a matter of fact I was extremely interested to learn the information in the presented video and I do find the model in itself to be brilliant conceptually, however, every model has limitations.
@@sams7068 as a matter of fact Igor Rausis never scored particularly high within the model, the give away was the consistency. He got banned and stripped of his GM title because he got caught with a phone in the toilet analysing his position. Up to that moment FIDE did nothing and the investigation only started after he was caught. He might have been under suspicion for ages, but I will tell you for a fact many players were speculating he was cheating before FIDE caught onto it and the guy got away with it for at least 4 years.
@@clementodenknirps7161 There is no such thing as "easy to find" mathematically.
I like presentations like this where you present a set of facts and leave it to the audience to make their call rather than some prominent GMs people who make deductions thereby biasing their audience.
No..
It's a biased an illogical with wrong conclusion and way of deduction,
With clear logical mistakes in the analysis
This video deserves more views. Nice and interesting analysis. Hopefully more players will get a better understanding of the anticheating methods and also the difference of perfoming high in a tournament or a game based on rating performance vs a more objective approach and analysis of what it really means to perform better when you put the games under the microscope and computer analysis combined with a sophisticated approach of discarding forced moves and opening moves.
Dont worry, its gonna blow up.
I’m very glad I found this channel
The whole idea behind Regan's system depends on knowing the actual playing strength of a player. For non-cheaters, you do know this, it is their rating (with a small to moderate +/- because of form, age etc.) For cheaters, you have no idea of their actual playing strength.
So you have a cheating detection system that is only good at catching the people who actually do not cheat.
It is still not a useless system, because it can detect someone starting to cheat.
The very obvious flaw in your take is that it assumes that Niemann cheated in most, if not all of his OTB games. That's a pretty bold claim.
@@pelic9608
Please qoute the passage where I made such a claim.
@@japphan "For cheaters, you have no idea of their actual playing strength" implies that in general terms for the specific case of Niemann.
@@pelic9608
No.
My statement is about the anti-cheat system, and one of its drawbacks.
Sure, it applies to Niemann. And Carlsen. And Dufus. And God.
The difference between player’s rating and the computer analysis is that rating only takes into account actual outcomes of the games against other rated players. The analysis looks at the actual moves in every game and compares it to the model corresponding to each player’s strength. Cheater would have to cheat very consistently or at every move of every game to make sure they are not caught. He would have to mix wins, losses and draws, make blunders and also finding winning moves only according to his rating. While this is possible, it is very hard to pull off especially over the longer periods of time.
But HOW do players chess players cheat OTB?
I don’t understand the ability for a transmission of next moves. If the game isn’t being streamed, or is but with a delay, how could anyone assist in the cheating?
Mobile phones in the washroom are the obvious culprit, because in smaller tournament, nobody is gonna check what you do in there.
But even on GM level, there are various way to cheat, such as your partner in crime might be simply visible to you and giving your coded gestures.
The idea that other players play worse against him rather than him playing better is a nonsensical statement. To play moves that your opponents find difficult to respond to is your better better play, not their worse play.
but not "absolute better", only "human" better
So with a lot of the recent games by Hans where he is winning in 28/32 moves... And all the opening/forced moves being ignored would there be enough moves to identify anything?
Also if a cheater is aware of this method of ROI can they not make sure they have bad days to balance it out?
That's the other thing. By Regan revealing the method of his analysis, an astute cheater can use that to devise a cheating strategy that would fit that model's definition of fairness.
How to calculate the ROI of an event? won games / played games?
It is move level (or an aggregate of move level, to be precise). Roughly like this: Take out all insignificant moves (forced, opening), calculate how large percentage you find a top move[1] and normalize (using skill level, I suppose ELO here) so that 50 becomes your expected value. I guess the problem is that most super-GMs seems to think (see Hikaru, Caruana, Carlsen) that even having a single binary signal during every game (like, "in this position there is a huge difference between moves and if you find the right, it is basically winning") would make them crush the field.
It is in fact quite similar to in poker. Obviously RTA (real-time assistance) is banned from online poker, but if you do it sometimes and selectively, it is basically impossible to detect, I would think.
I think finding only the crucial moves, maybe even in crucial games and doing statistics on those (maybe correlated with live/delayed brodacasts and with/without audience) would be interesting. My gut feeling is that this drama will be dissolved with data from Chess.com, though. No OTB cheating proved, but a much larger sample with super likely cheating and thus credibility of Niemann drops dramatically and he'll not be invited "anywhere". Sorry to say, but there is actually only a very small chance that he cheated only exactly when caught, which he sort of claimed in that otherwise seemingly trustworthy interview after the SC... He has also been awfully quiet after Chess.com basically doubled down and said there was more to it than he admitted.
All of that being said, super interesting analysis and shows that it is not clear cut and easy to see here. Also easy to cherry-pick data, have not yet seen qualified opinions on this video th-cam.com/video/jfPzUgzrOcQ/w-d-xo.html showing some very high correlations with engine for particular Hans games.
[1] - I am not a good chess player (enjoy the game a lot though) and my statistics skills are rusty, but I guess computing this based on centipawn loss would be the straightforward route.
In casual, non-tournament play, people were blowing Hans out of the water. He admits he cheated 4 years ago & 8 years ago. A leopard doesn't change his spots.
Right, but what's his expertise in vibrating beads?
Well, a cheater doesn't have to use the top engine move very often. For example in a position where the best move gives evaluation +5.3 the second best move gives evaluation +3.5 and the third best move gives evaluation +2.3 he may as well play the third best move [which will be considrered an error of magnitde 3 (because 5.3-2.3=3) and win as well]. Also in a position with great complexity he may also from time to time select a move that gives evaluation +0.7 and regardless win the game if he opponents fail to play perfectly due to the high complexity of the position. If he uses suboptimal moves that win regularly, being a "smart cheater" then the analysts will be unable to detect with certainty the cheat.
I personally believe that Hans Niemann is a cheater. The main reason that I believe that about Hans is that, unlike any other 2600+ player, Hans is unable to provide a consistent analysis of a position in postgame interview providing lines and variations.
Kinda late, but as i understand, playing top engine moves is worth it when you know how to capitalise on them. By that logic, Hans couldn't get away with them, 'cause he can't analyse at the same depth as engines and make following moves that engine thinks would give you the advantage, i guess
Would it be a fallacy? Because if he cheated before, he gained certain points of Elo. Further analysis will be applied on his cheated Elo. So the ROI makes no sense, if he have cheated at some games. Or am I wrong?
it would make his performance way worse unless he was using it all the time - if he wasnt capable of actually reaching grandmaster level it wouldnt really matter if you cheat sometimes for ROI as it would bomb during the games you werent cheating in due to the inflated rating
Excellent remark!... Could a mathematician enlighten us on this point?
@@fredericpelloud7536 think of it like this ROI is essentially your average expected performance for your rating on a bell curve with 50 being your rating and 100 being playing so unbelievably far above your rating youre like a 1500 beating a 4000 rated engine. If neimann truly has a lesser level of skill than his opponents, when he doesnt cheat (to avoid suspicion) he would immediately become severely less accurate and his ROI would become suspiciously low. Also online chess (where hans cheated) doesnt actually matter for FIDE rating afaik?
@@youareaspook5897 Where Hans *admitted* to cheating.
If you only cheat occasionally, you're far more likely to get away with it. That's where Igors went wrong. Also, if you do worse performances elsewhere on purpose, you can cause it to average out and make it look more normal. You're completely ignoring the abnormal peaks.
How do you arrive at the conclusion that there are abnormal peaks?
@@Sarah-oj7bh nobody else has comparable engine stats to his supposed peaks. Literally not a single person in the world. Not even remotely close. It's obviously abnormal.
@@apocalypseap so you're referring to Yosha's video? So can you explain to me what this engine stat means?
crazy how his rating slowly and steadily increased over time
I thought it was weird too.
I don't know--I still trust Magnus's OTB instinct when he felt that in their St. Louis game Hans didn't spend much time concentrating on critical positions, and beat him in an endgame in a way only a few players could (Magnus would know, since he's an endgame wiz). You can't quantify this feeling. But I think the World Champ can "sense" if someone's play is a bit fishy, over-the-board, just like how a strong master disguised as a weak player plays a street hustler, and immediately the street player notices they are not that weak. You just know, based on the moves chosen, and the time taken over the moves, the hesitation, the expressions, and other "human" elements.
Lmao that sounds stupid
There weren’t many critical none forced positions
@@tajneeley Sure there was! Did you even go over their game? The endgame had all sorts of tricky pitfalls and traps, positions where Magnus could usually squiggle out of, due to his superb endgame maneuvering. But Hans kept playing the best, strongest move each time, pinning Magnus down. That's probably why he got suspicious, as he said. If you think that sounds stupid, then you think Magnus is stupid.
Yoooo I have always believed in due process. Thank you for the analysis and I await a response from Niemann defending himself
he doesn't have to defend himself, they didn't even provide a single proof against him, that's like they are barking dogs.
tell me, would you talk to a dog if he starts barking ?
@@pythonwolf3817 When the dog is the world champion, then you better respond.
@@stevenclark5173 sorry man, that's not me, even if it's the world champion, to me, the dog is a dog
It's a biased an illogical with wrong conclusion and way of deduction,
With clear logical mistakes in the analysis
Appreciate the work put into your videos. I really like this channel.
I'm glad to see an expert concluded that Niemann was not cheating.
For those who still think he was cheating, I have to ask: "How did he receive the move information from the engine??"
It's great for catching consistent cheaters. Not so great for catching opportunistic ones. If you pick your moments, and play without cheating a fair bit of the time, you'd have a normal range of variance in performance still. It's not hard to defeat a summary analysis which is looking for averages and a regular range. You really want a model to detect an anomalous move, not a consistent over performance these days. If you play most of the game at your standard level, but then on a crucial move, get insight, and then continue playing at your level, it wouldn't show up on an analysis like this, it's comparing your overall performance.
The problem is that you can always play moves that are "too good" just because you overlooked some other human moves. The only way you can detect such a thing by pure analysis of the game (i.e not catching the guy red-handed) is to see discrepancies on a larger number of matches. The data is showing that Hans's games can't be distinguished from the game of a good player.
@@desarguesbaptiste5577 Which is exactly the point: you can't catch a clever cheater through pure analysis.
@@desarguesbaptiste5577 the data isn't even showing anything here, because it's from individual tournaments which averages out instead of being shown from individual games. Now that's the data I'd be really curious to see since suspicious games would look really anomalous
Great logical deduction! absolutely agree and shocked by the illogical conclusions of this video.
How can they not see what you've just said?
"
10:16
I absolutely disagree! and think you are biased.
It would absolutely give you the result you want:
More rating, more money and fame and without the risk of being caught.
It's the move of the "very smart cheater".
You don't have to win the tournament to improve your financial state.
As long as you have a better rating than you could have had - then you are improved.
"
Why does the 7.5/9 tournament display 48% ROI in the table and then 50% in the tournament summary?
If Magnus says he’s cheating then he’s cheating. I would take Magnus’ words who was seating right in front of Niemann for several hours during the live game over some expert who studied his games over brief video clips and computer analysis.
This is such a disgusting blind follower comment. That represents everything wrong with online mobs.
If you are being serious: this is hilarious
If you are trolling: you’re not funny
World chess champion has somehow found a way to be a master of body language analysis, a field in which the methods are so inconsistent and unreliable that it’s relegated to the arena of pop psychology pseudoscience
Just one thing tho, besting the current World Champion (or any sufficiently renowned top player for that matter) would give quite a reputation. It may give the impression of "an underdog that defeated top 1 on multiple occasions". Metaphorically speaking, it's not about an immortal bleeding, but for a mortal to make an immortal bleed.
I still remain neutral at this drama, but there's no doubt this will make Niemann either famous or infamous.
If he came on a better light after the drama (Cheating not proven, Carlsen acknowledging him,etc.), then he'll be famous for standing up to all of that pressure, one of them from the World Champion himself.
If he came on a bad end of the drama (sufficient enough probability of cheating or something along those lines, but still not enough to ban him from playing) then his infamousness would be a good psychological card. Imagine playing against a probable engine or just a mere human. You would even doubt if they blunders.
I think the obvious conclusion here is that Hans isn't cheating, he's just blackmailing his opponents to play worse. Magnus didn't drop out because he suspected cheating; he dropped out to make sure his family was OK.
Lmao chess mafia 🤣
Why sort on ROI and not time, what if he plays five tournaments in succession with a ROI of 60, that would be suspicious right? or after how many tournaments would it be suspicious?
This was really good. I enjoyed. I hope you will come up with more videos like this.
he played badly when there wasnt live transmission...
He also played very good during such. Where were you?
1) If your opponent plays very well, is it not expected that the performance of the opponent (Opponent's figure) will automatically be lower?
2) Don't you need a baseline to generate the 'Opponent's Figures'. For example, i) the statistics (Opponents figures) of GMs playing blind against a chess engine and ii) the statistics (Opponents figures) of GMs playing blind against a chess engine where a human is purposely adding fake moves in order to try and confuse chess cheating detection software?
3) Knowing that only certain moves are used in the statistical analysis, means that a chess player can use cheat moves sparingly in the middle game, but use cheating to make perfect opening and endgame moves?
4) Recognizing forcing moves may not be obvious, so cheating to recognize forcing moves would not be counted in the statistics as people assume that every chess player will see the forcing moves without the need to cheat?
Innocent till proven guilty (and everyone benefits from that and you won’t appreciate that unless you are wrongfully accused for something, not just chess). Hans is probably using unorthodox moves or psychologically impacting others in unexpected ways combined with having significant growth in certain aspects of his game (making him erratic to some) as his game improves overall. Think of a teenage tennis player who has worked tremendously on his topsin and slice serves for a few months, improving greatly in those areas (while improving at a more regular rate in his overall skills) and perhaps who has studied the games of a few of the people better than him (i.e. Magnus’ games in Hans’ case). And it’s possible he accidentally or intentionally found out about Magnus’ plans in one game.
Has he admittedly cheated in the past online? Yes. Did he cheat more times than admitted? Possibly. We’re also dealing with young teenagers learning about themselves, life and the “philosophical” rules in general. He’s clearly a bit of a rebel aka independent thinker. Give him the benefit of the doubt OR trust and verify. Let’s see how he fares in the coming months and years.
And as intelligent/skilled as Magnus is at chess and even though he has been right before, he could be wrong. Many elite and brilliant CEOs, surgeons, pilots, etc. have made catastrophic errors in judgment and have been wrong to the peril or demise of many others.
"Innocent till proven guilty" He admitted to him cheating twice when he was caught. So, guilty when was caught and innocent when not. "He’s clearly a bit of a rebel aka independent thinker." Because of him cheating online? LOL But Niemann cheating (verified in the past) will haunt him.
@@mytube2011z yeah but u got no proof he cheated this time, if you do share it if not shut up.
@@mytube2011z If he is still there playing, that means he was not perm banned. So we have to live with that. Innocent until proven guilty for now. If that isn't enough, they should make new rules to prevent anyone who has any historic of cheating for now on.
he is a cheater and has cheated, guilty until proven innocent because of this.
@@pony6 you are a liar because you have lied before, guess whenever you get accused of something it's guilty until proven innocent?
So is it true that Regan's method can be completely circumvented, by tossing a coin before each game to decide whether to cheat or not?
Han’s opponents in crucial games may have been flawed by his speed of calculation during ambiguous positions. A deeper insight into his facial expressions throughout questionable games may shed some light.
What do the odds in Regan's models say about Magnus misjudging a known cheater having just played actual matches with him? All we have to do is input all the other times Carlsen has resigned a match, quit a tourney, and publicly commented with an accusation, and... let's just put that data into the formula and... oh, wait...
So we have a problem now, as there has been shown that Hans has played more than 20 games with 100% engine correlation. Magnus has palyed 2 of these games in his whole career, Bobby Fischer not a singe one and his engine correlation was 84% during his winning streak.
Basically there`s strong statistical evidence now, that Hans has used an engine to cheat in chess.
Seems like this analysis here is wrong. Time to reevaluate the method.
It seems like only a matter of time before its proven. Thats the problem with cheating in the modern age. Every game is logged and can be analyzed and compared. People say "there is no evidence". But I would say, if this were a court of law and you presented a jury with his engine correlation statistics they would probably deliver a guilty verdict. What possible defense is there for that? He just randomly morphs into the greatest chess player of all time?
LOL because he's not been found guilty by this analysis it's "time to revaluate the method"?... You would be good as a court judge :)
i dont agree with this theory, can a cheater become better at cheating so that he wont get caught ? yes ! notice as his ROI improves so does his opponent, this means he is adjusting the level of cheating according his opponent. He is smart not to go full Stockfish that would be pretty obvious, he is a prepared cheater and throws in random moves ignoring the feedback from Stockfish sometimes. Intrestingly this is EXACTLY what chesscom had released of Niemann's online chess games and their Antii-Cheat detection, which isnt fully expĺained how that exactly works but is does describe a part of it's functionallity that it reckognizes someone using every available attack or defense from the chess engine. As some experts pointed out that not even the best chess player in the world can memorize every strategy that was ever written in a book. And Hans Niemann when asked about his moves didnt gave any correct anwer besides his incoherent rambling.