To the best of my memory where it folded the king flush to my ace flush I raised PF and he called, flop was check/check I bet turn and he called, river he bet and I raised him and he folded. I would bet half of my bankroll here that if the flush misses he bets me out.

In 2nd hand he raised pf and I called, flop I bet, he raised I reraised. On turn I check raised him and he folded.

Third hand had a cap pf, flop and turn were check/call and on river he opened and I raised him and he folded. He has a monster in this hand. He can only lose to 44, 88, 99, TT, JJ, KK, AA or Qx. I'm sure there are a certain % of people that would try to bluff with the board there if they had not played this machine much. Folding there is even worse than the AA hand IMO.

On the hand the hand I cannot get to show I raised pf with 9 suited and he called with T9. We checked it down until river when I opened and he called me with unimproved ten high.

I have a collection of bizarro hands, this is just a sample, him folding the 2nd nuts is by far the most disturbing. He has laid down quite a few boats. In one instance he check folded 22 into tripped flop. Every one of his big hand laydowns were right.

As for the set of aces laid down on the turn there are few people that aren't going to call that down heads up in limit anyway and obviously nobody lays it down on the turn getting 11-1. Considering he virtually never lays down an unimproved ace or king, even to a check raise, it makes that hand even more disturbing.

I stopped playing the machine because he played so terrible but yet slaughtered everyone. I had a nice run on it when they first put the machines in but gave most of it back. One thing that always made me wonder was I would see the guy that seemed to have invented the machine playing it at Red Rock. Tell me that isn't strange.

Does anyone have any information on the shuffling algorithm used here? I have a theory, but it depends on there being a big error there. Would gaming go through and check that there is sufficient entropy in their deck of cards? Other people have screwed this up in the past.

Yeah, that's effed up.

Am I the only one that thinks that's not that unreasonable?

A flop checkback range will be Ace heavy.
A turn betting range after a flop checkback will be spade heavy (not many pairs).
A river raising range won't include many Js since the bots check/calling range has many Ks and As in it. 9s is even more discounted because of the flop checkback.

Therefore a river raising range is very polarized and the pot isn't very big. So if the bot figures a gto range won't bluffraise the river often enough (1/8 times or more I think?) it can bet/fold.

Edit:
the way you described the hand before it sounded like the bot had called a turn bet and then check/folded the river, which would obviously have been insane.

Based on the screen shots, it's clear that your memory on how the betting went is a little off. In all three hands, the game appears to be 2-4 holdem. On the first hand, the screen shot shows that you put $22 into the pot but your description accounts for only $16. In the second hand, the screen shot shows you put in $24 but your description accounts for only $18. In the third hand, the screen shot shows you put in $24 but your description accounts for only $22 (missing a small bet).

It will give up some hands... so you might think you have a chance/feeling of being able to beat it. I have a feeling the machine is looking at as the amusement!

Is there any way to determine if the game is rigged or not? It's either rigged or the experts who can beat it are not bragging about their exploits in this thread.
As far as I know, this game has not appeared outside Vegas. (And I hope the beastly contraption stays in Sin City.)I wonder why--as reported by other posters--some of the games were moved offline in some of the casinos?

If any other slot machine went missing for a month there wouldn't be conspiracy theories abound. Pokeraddict links sketchy screenshots of the bot folding the 48th nuts to a river raise and then another screenshot where the bot must be rigged because it folds top set on a 4 flush board despite getting immediate odds to continue to the river. Wait, what? If the bot were rigged why would it make a clear mathematical mistake to its own detriment? And regarding the K flush hand, for every time I see the bot hero folding the river, and trust me, I consider folding 3rd pair a hero fold for this thing, I have dozens of other instances where it calls with unimproved high card garbage on coordinated boards where it doesn't even have proper pot odds to call against a completely random hand. Occam's razor people. I find it far more likely that there's just programming flaws with the brand new game than the explanation that there's some elaborate conspiracy afoot.

Its logic is just bad sometimes. I remember being properly startled when I value bet an 8 on QTJ95 rainbow and then having it just call with a K. And its decision making in capped pots is just atrocious. Despite its apparent timidness that it's up against AK in the aforementioned pot, it feels perfectly comfortable 4-betting the river against me on AJTTK rainbow with a naked Q, perhaps that's an example of one of the personality shifts Sklansky mentioned. Speaking of which, these personality shifts (specifically its 3-betting starting hand selection and frequency, but also some postflop decisions like its donking tendencies) are really apparent once you play against it enough, but I highly doubt they were implemented so that the bot counter adjusts to the player because it will exhibit the same tendencies for hours at a time. I find it much more likely they were implemented to make it harder to beat the game in general so people can't just author a strategy guide and be done with it. That's not to say that I don't think there's still exploitable tendencies that are universal, or nearly universal, across its play, because there definitely are. For example its turn (and arguably, river) 3-betting range is exceedingly tight and worse still is that it's seemingly unaware that the player is (or should be) aware of that fact and therefore overvalues hands like 2 pair in capped pots. It's also too passive in general when the player 3-bets preflop which is strange because it has no problem staying aggro when the player 4-bets with a presumably tighter range and position. On that note I think people greatly exaggerate the skill level of the bot.

I find it almost comical when people try to dissect the bot's calldowns by spouting off terms like range polarization that clearly have no application in its decision-making (not to mention the fact that HU limit is probably the most unpolarized form of poker on earth and value betting two pair on a 4 flush board is standard, so the moment someone even mentions it I just instantly dismiss their post as pedantic diarrhea). Its exceedingly bad decision-making in capped pots makes it obvious that the bot does not use deductive reasoning. I believe it was mentioned elsewhere that it makes decisions off empirical experience over a very large sample, which makes lots of sense to me and that's why it will play fine 99% of the time and then spaz something stupid the other 1% of the time that it encounters a situation that it's not as experienced handling. Given the extremely large different combinations of hole cards, flops, turns, rivers and every player and AI decision up to that point, it makes perfect sense that it would encounter spots it hasn't been in before and sometimes make obviously incorrect plays. That's all there is to it.

Interesting stuff. So how much you beating it for?

I don't really know, I honestly wouldn't even be able to guess just because it was comingled with all the other poker money I was using over several months, but it felt like I was beating it for a pretty healthy clip over at least a 10K hand sample. There's a lot of advantages you can leverage as the player against the machine, some are obvious from a theory standpoint (for example, you can 3-bet preflop strictly for value), some are specific to the bot having extremely well-defined ranges in certain spots (I'd rather not discuss these, at least not publicly). It'll c-bet dry boards with the best of them and because it can play standard spots competently it will easily crush most opponents (I saw one guy playing $20/$40 at the Venetian open fold QTo to a preflop 3-bet). It just seems extremely weak in 3-bet and capped pots, and because most competent limit players can play standard spots just as well, I think beating it is easier than most people think.

I'd be very happy to book the machine vs you if we can figure out a way to verify the accuracy of your results.

What kind of sample are you looking for? Unless you're trying to get an absurd number of hands tracking results shouldn't be difficult. What odds are you laying? I'm not in Vegas anymore, but would be interested.

Lol you want me to lay you odds? I'd obviously need a third source I trusted to do any bet, and I'd need some time to consider terms but prolly something like x*result.

when i was in vegas a lot of people were bragging about how they were beating this thing over x amount of hours for y amount of bets and i couldn't help but laugh inside. they have no clue just how tough this thing is. at best i think even experts would only have a slight edge

that being said, i beat it for x amount of hours for y amount of bets =)

i will say though that i think it is pretty tough. it does weird stuff and bluffed me off some pots i thought were ridiculous spots ie sick soul reads that had me insta-quit it a couple times because i was so steamed

I'd book the machine against anyone in the world.