No limit holden video poker machine - optimal strategy
This looks very interesting, if not a bit scary (Re: AI).
If a NO-LIMIT version of that existed, this would create an opportunity for a simple math based short-stacking & rat-holing approach, eliminating all post-flop implied odds because you can buy-in very short, cash-out, and re-buy as a "new" player as much as you like. From investigation of this HU-LHE machine for a thread with some folks on a poker-oriented discussion site, I believe the machine plays very close to game theory optimal, and is a very difficult opponent by really "playing poker" with a conventional sized buy-in allowing for multi-street play. I confirmed to my own satisfaction statistically (after some attempts to look at it mathematically) that doing this short-buy on the limit HU machines produced very close to break-even results, but doing the same in a NLHE structure would be very different, and worth investigating whether it would be possibly profitable.
Quote: AsswhoopermcdaddyCan anyone comment on what optimal strategy is supposed to be for the no limit holden video poker machine?
uhhhh.... mixed-strategy nash equilibrium?
http://en.wikipedia.org/wiki/Mixed_strategy#Mixed_strategy
I was thinking the same thing. I don't think they can make a (legal) good enough NL machine.Quote: DrawingDeadAre you sure this isn't the LIMIT hold 'em head-up game?
Quote: DrawingDeadIf a NO-LIMIT version of that existed, this would create an opportunity for a simple math based short-stacking & rat-holing approach, eliminating all post-flop implied odds because you can buy-in very short, cash-out, and re-buy as a "new" player as much as you like. From investigation of this HU-LHE machine for a thread with some folks on a poker-oriented discussion site, I believe the machine plays very close to game theory optimal, and is a very difficult opponent by really "playing poker" with a conventional sized buy-in allowing for multi-street play. I confirmed to my own satisfaction statistically (after some attempts to look at it mathematically) that doing this short-buy on the limit HU machines produced very close to break-even results, but doing the same in a NLHE structure would be very different, and worth investigating whether it would be possibly profitable.
There is no mathematical advantage to being short-stacked in heads-up play. Heads-up, both stacks are essentially of equal size. Being short-stacked only confers a mathematical advantage in games with 3 or more players.
Quote: AsswhoopermcdaddyIf I remember correctly it was indeed limit, but I noticed that the reraises increased progressively so it felt like no limit......so optimal strategy is to buy in with a low stack?
The term "no limit" refers to being able to bet as much as you want on a single bet, not on the cap (or lack thereof) of the number of bets on a betting round.
If you don't know this, I'm guessing that poker is not your game, so I'd suggest staying away from these machines.
Nope because I think if you run out of credits you lose the button. And since he always gets the button first. You cant one hand it every time.Quote: AsswhoopermcdaddyIf I remember correctly it was indeed limit, but I noticed that the re raises increased progressively so it felt like no limit......so optimal strategy is to buy in with a low stack?
I find them very fascinating its to bad they didn't take off better. I heard they have a second version?
Unfortunately I think they crush normal players to fast. People, even some AP's, think the machines cheat, If AP's think this then certainly every ploppie thinks this.
At the beginning they had a glitch in the programming. It played weak with a certain system. This didn't last long.
I'm up over all on these however I don't have the Patients to really put significant time in. I'm also wary and I don't trust something about them,not sure what, I cant put my finger on it. I
As far as I can tell NO ONE can beat them constantly. Guys I talked to always start off with a bang and then slowly they get crushed. I have a feeling this is due to patterns they think they see.
I have talked to a few very smart (mathematically, analytically and logical) AP's who play both poker and machines and understand both very very well. They wont touch them anymore now that most all the theoretical has been cut.
I have heard big name pros say they can beat them(guys who helped promote them?) but they have some excuse why they never play. One problem with the big limits (unless they fixed this) A hand pay resets the button to the Bot every time.
Does anyone know all the locations of them and if any still have a good rating? I wouldn't mind having a go at them if I can get some descant benefits.
Quote: AxelWolfNope because I think if you run out of credits you lose the button. And since he always gets the button first. You cant one hand it every time.
I find them very fascinating its to bad they didn't take off better. I heard they have a second version?
Unfortunately I think they crush normal players to fast. People, even some AP's, think the machines cheat, If AP's think this then certainly every ploppie thinks this.
At the beginning they had a glitch in the programming. It played weak with a certain system. This didn't last long.
I'm up over all on these however I don't have the Patients to really put significant time in. I'm also wary and I don't trust something about them,not sure what, I cant put my finger on it. I
As far as I can tell NO ONE can beat them constantly. Guys I talked to always start off with a bang and then slowly they get crushed. I have a feeling this is due to patterns they think they see.
I have talked to a few very smart (mathematically, analytically and logical) AP's who play both poker and machines and understand both very very well. They wont touch them anymore now that most all the theoretical has been cut.
I have heard big name pros say they can beat them(guys who helped promote them?) but they have some excuse why they never play. One problem with the big limits (unless they fixed this) A hand pay resets the button to the Bot every time.
Does anyone know all the locations of them and if any still have a good rating? I wouldn't mind having a go at them if I can get some descant benefits.
It is entirely possible that they are playing a perfect mixed strategy (I assume that it's possible; I don't really follow poker research so I don't know if this is a solved problem). If they are then the game is not beatable long-term; the best you can do is break even if you also play perfectly (which is realistically not possible without a computer).
Even if they are playing a very good game theoretic strategy (ie, not perfect, but close) it's highly unlikely that a human could do better without examining the programming and finding the flaws in their strategy. Even then, I would not like a person's chances.
Quote: AxelWolf I have heard big name pros say they can beat them(guys who helped promote them?) but they have some excuse why they never play. One problem with the big limits (unless they fixed this) A hand pay resets the button to the Bot every time. [/q
There was a lot of scuttlebutt online about this game when it first came out. And there was a glitch that got exploited until it was fixed. I think the excuse the big name pros used when they quit playing was they took out the 10-20, 20-40, and left only lower betting like 2-4, 3-6, etc."Quit trying your luck and start trying your skill." Mickey Crimm
I agree *IF* preflop ranges are the same for each opponent. The size of the shorter stack becomes effective stack size which is in-play for both. What short-stacking does is eliminate implied odds from post flop play, and *IF* pre-flop ranges are the same for the human and the HAL9000 Pokah-Wizard thingamajjiger potentially makes the LHE game close to a break-even slot machine, which could be interesting if it provided comps/credits at some rate that was material, which it did not. But "optimal" on this machine may be a long-range expectation of a loss as close to zero as possible.Quote: AxiomOfChoiceThere is no mathematical advantage to being short-stacked in heads-up play. Heads-up, both stacks are essentially of equal size. Being short-stacked only confers a mathematical advantage in games with 3 or more players.
My experience was that I think some pretty decent live poker players had a reaction of: "No f'ng way; this mudderforking gizmo just has to be rigged!" (close to a direct quote) when trying it because correct head-up ranges are so very wide, much more so than most who play full-ring games would imagine, including me. I didn't really understand this until I began researching what should be my optimal range for experimenting with this thing. And what I found is that it should never open-limp (not so surprising) and should also fold very few hands, resulting in raises with hands that makes it appear (to many full-ring game players) to be a maniac chimpanzee on crack (or rigged) when it beats you like a mule with "that crap."
Basically I don't think you will beat it, at least in the LHE version. You could very nearly break-even. Essentially I did. While carefully recording everything it did for several weeks, at the multiple locations where they were installed, which is a lot of hands because it plays fast, I think I luckboxed my way into taking them for dollars. Not thousands, or hundreds. or tens of dollars, but a final total of single digit dollars, resulting in a win rate of essentially zero, and what I believe was actually a long range expectation of a small fraction less than zero. But I (and the others who were interested) were only considering the LHE machine that existed.
I believe the list of locations I posted was complete as of that time, and I think the availability in the Las Vegas area has more likely contracted rather than expanded since then.
Quote: DrawingDeadDisclaimer: I would describe myself as merely a moderately competent only slightly better than mediocre poker player, who normally relies upon his ability to only play with tables of drunks, lunatics, and drooling imbeciles.I agree *IF* preflop ranges are the same for each opponent. The size of the sorter stack becomes effective stack size which is in-play for both. What short-stacking does is eliminate implied odds from post flop play, and *IF* pre-flop ranges are the same for the human and the HAL9000 Pokah-Wizard thingamajjiger potentially makes the LHE game close to a break-even slot machine, which could be interesting if it provided comps/credits at some rate that was material, which it did not. But "optimal" on this machine may be a long-range expectation of a loss as close to zero as possible.
If I understand you correctly, you're saying that if you play sub-optimally, you can reduce the cost of your mistakes by being short-stacked. That's probably true.
However, it will never give you an advantage, and, if possible, you would be better off just correcting the errors in your play.
I was mostly replying to this:
Quote: DrawingDeadbut doing the same in a NLHE structure would be very different, and worth investigating whether it would be possibly profitable.
My point is that it can't possibly be profitable.
That is true if there are identical pre-flop ranges. If pre-flop ranges differ, and effective stacks do not allow for post-flop play, it becomes a matter of simple math with the difference in the ranges determining the long-range outcomes.Quote: AxiomOfChoiceIf I understand you correctly, you're saying that if you play sub-optimally, you can reduce the cost of your mistakes by being short-stacked. That's probably true.
However, it will never give you an advantage, and, if possible, you would be better off just correcting the errors in your play.
I was mostly replying to this:
My point is that it can't possibly be profitable.
ADDENDUM: As an aside, I think all humans play sub-optimally. To varying degrees.
Quote: DrawingDeadThat is true if there are identical pre-flop ranges. If pre-flop ranges differ, and effective stacks do not allow for post-flop play, it becomes a matter of simple math with the difference in the ranges determining the long-range outcomes.
The assumption is that the computer is playing the correct pre-flop range. You can break even with it by playing the same one, or lose money by playing worse.
With that assumption, which I believe is true for the LHE machine that exists, we fully agree. Would a hypothetical NLHE version of HAL9000 on the casino floor adjust to play one sort of game with deep stacks and large post-flop implied odds, and adjust to another with a pre-flop "all-in or fold" game? I haven't spent time considering the matter; perhaps it would end in the same place.Quote: AxiomOfChoiceThe assumption is that the computer is playing the correct pre-flop range. You can break even with it by playing the same one, or lose money by playing worse.
Quote: DrawingDeadWith that assumption, which I believe is true for the LHE machine that exists, we fully agree. Would a hypothetical NLHE version of HAL9000 on the casino floor adjust to play one sort of game with deep stacks and large post-flop implied odds, and adjust to another with a pre-flop "all-in or fold" game? I haven't spent time considering the matter; perhaps it would end in the same place.
I have no doubt that the correct strategy depends on stack size as one of its parameters. I have no idea how far along research in this field is, so I don't know how feasible this is. IIRC a lot of research was being done at the university of calgary.
Quote: mickeycrimmQuote: AxelWolf I have heard big name pros say they can beat them(guys who helped promote them?) but they have some excuse why they never play. One problem with the big limits (unless they fixed this) A hand pay resets the button to the Bot every time. [/q
There was a lot of scuttlebutt online about this game when it first came out. And there was a glitch that got exploited until it was fixed. I think the excuse the big name pros used when they quit playing was they took out the 10-20, 20-40, and left only lower betting like 2-4, 3-6, etc.Suck dope, watch TV, make up stuff, be somebody on the internet.
Quote: DrawingDeadI'm in Aria fairly often (though not to play these things) and I noticed as recently as a few months ago the ones at that property still offered the higher stakes. So, if that is someone's excuse, it is lame. They are right in front of one of the 3 biggest busiest most prominent poker venues in town. I rarely see anyone playing them.
I was in Aria yesterday, and remembered this thread as I walked past these HU-LHE machines (photo by me 6/28/14):
I don't recall exactly what the full range of stakes was in the past at this property, but I think it had at least one additional level as an option, like probably 20/40L.
On the other hand, I'll still say that if some of the folks in the most buffoonish pompous poser end of the local poker community's pond say they don't bother with it SOLELY for THIS reason, I still think that sounds pretty lame. I found the play on this thing tends to be very fast with an extraordinary hand count compared to anything else I've experienced, unless you're working really hard to diddle & dwaddle beyond all sensible reason & take extraordinary care to carefully inspect the assets of all the cocktail servers to keep it as slow as you can. So playing 10/20L head-up vs. this thing can quickly amount to a lot more than chump change, in my opinion.*
On yet again another hand, as I was saying to Phil Ivey, Ghandi, the Pope, and the Queen of England just the other day, I would've taken a seat in the 200/400 mixed game to vacuum up every last chip on the table while I was in the neighborhood, but unfortunately I suddenly remembered that I'm overdue to scrub the floor underneath my fridge, so I regret that I'll just have to catch it next time.
ADDENDUM: Re-reading this I think the way I wrote that might not be clear. I am NOT intending to say that "a lot more than chump change" will likely be won rather than lost, just that significant enough amounts can be played through it (at some locations) and risked on the thing if that's what someone is after. This machine is a very tough opponent. If it had legs and took a seat next to me I'd leave that table.
Quote: DrawingDeadI think I was at least partly wrong about this:
I was in Aria yesterday, and remembered this thread as I walked past these HU-LHE machines (photo by me 6/28/14):
I don't recall exactly what the full range of stakes was in the past at this property, but I think it had at least one additional level as an option, like probably 20/40L.
On the other hand, I'll still say that if some of the folks in the most buffoonish pompous poser end of the local poker community's pond say they don't bother with it SOLELY for THIS reason, I still think that sounds pretty lame. I found the play on this thing tends to be very fast with an extraordinary hand count compared to anything else I've experienced, unless you're working really hard to diddle & dwaddle beyond all sensible reason & take extraordinary care to carefully inspect the assets of all the cocktail servers to keep it as slow as you can. So playing 10/20L head-up vs. this thing can quickly amount to a lot more than chump change, in my opinion.
On yet again another hand, as I was saying to Phil Ivey, Ghandi, the Pope, and the Queen of England just the other day, I would've taken a seat in the 200/400 mixed game to vacuum up every last chip on the table while I was in the neighborhood, but unfortunately I suddenly remembered that I'm overdue to scrub the floor underneath my fridge, so I regret that I'll just have to catch it next time.
If all the card benefits have been cut from these machines then I think that speaks volumes about who is winning.
I wondered about that too. But then a peculiar thing (at least from how it struck me with my general lack of knowledge & experience with machine games) is how inconsistent they are. Red Rock counted the "coin-in" for comps & credits in their system exactly like they do on the the regular VP machines they have all over the casino floor at Station properties. Meanwhile at the opposite end of the range for these things in the Las Vegas area, at the MGM properties that have some here is the complete list of what you can get on your card from this game: "No." The variation made me wonder if they have any clearly defined idea of what they should want them to be worth... possibly related to the extremely limited number of these each of them has and the small volume of play on them from which to cook up a model to base that kind of thing on? I'm just going off into wild speculation here, and I freely admit I'm not well informed on how things work in that whole large electronic games part of the industry.Quote: mickeycrimmIf all the card benefits have been cut from these machines then I think that speaks volumes who is winning.
Quote: DrawingDeadI wondered about that too. But then a peculiar thing (at least from how it struck me with my general lack of knowledge & experience with machine games) is how inconsistent they are. Red Rock counted the "coin-in" for comps & credits in their system exactly like they do on the the regular VP machines they have all over the casino floor at Station properties. Meanwhile at the opposite end of the range for these things in the Las Vegas area, at the MGM properties that have some here is the complete list of what you can get on your card from this game: "No." The variation made me wonder if they have any clearly defined idea of what they should want them to be worth... possibly related to the extremely limited number of these each of them has and the small volume of play on them from which to cook up a model to base that kind of thing on? I'm just going off into wild speculation here, and I freely admit I'm not well informed on how things work in that whole large electronic games part of the industry.
I would guess that the 2+2 forums thoroughly discussed this game and it's archived.
Quote: mickeycrimmI would guess that the 2+2 forums thoroughly discussed this game and it's archived.
http://forumserver.twoplustwo.com/29/news-views-gossip/new-poker-machine-rake-free-headsup-limit-holdem-895984/
I'm sure that's true. I'm equally sure that slogging through a 2+2 thread for most adults (certainly for me and nearly everyone I know) tends to resemble swimming through a septic tank to snag the valuable nuggets that are scattered & mixed in it somewhere within all the foul smelling slimy dreck and the 14 y/o boys raised by wolves doing what 14 y/o cubs think will make them sound cool online. But I'm sure you're right that if someone cares about these things enough to do it, the relevant information would almost certainly be in there somewhere, and that'd be the place if they are up to doing the digging & sifting.Quote: mickeycrimmI would guess that the 2+2 forums thoroughly discussed this game and it's archived.
Quote: mickeycrimmI just talked to Al about the game. He played on two machines at Montbleu in Lake Tahoe. He had a friend that did the same. Al showed a 0.11% win after 200,000 hands. His friend showed a 0.05% loss. Al said you have to have solid game theory and know how to mix it up to hold your own in the game. The money was in the drawings and kickback checks. Montbleu cut the card benefits on the machines to nothing.
Quote: DrawingDeadThanks for taking the trouble to dig out the link. I can't believe I just went and ate the whoooole thing (sometimes skimmed through parts). I'm surely a sick man. Definitely now if not before. It may take me a week to decide for myself what I may have learned from it, but for the moment I get at least as much from reading your concise report of the personal experience of some folks you know. Though in fairness to the 2+2 thing, some of that is because I don't have all the background to digest some of the side discussions that make up a lot of the 2+2 thread, and also because it started as a breaking news sort of thing and I've already been introduced to Mr. Electro Poker Gizmo - and made a clumsy pass on his sister when invited in for lunch.
Al says the bot plays super aggressive and you have to do the same thing. The bot will call the turn and river with ace high so you get good value out of your pair hands. There was a programming error when the game first came out that some guys snapped to. I think Al said they created $56 tickets and played 10-20. You called the preflop raise, then called on the flop and turn, then you bet out on the river with less than a full bet left. The bot would automatically fold even the nuts.
All suggests you play 5000 hands at the lowest denom and see how you hold up.
http://www.nytimes.com/2013/09/08/magazine/poker-computer.html?pagewanted=1&_r=1
Thanks for posting it. After the first page I felt it was well worth saving so I can finish reading the rest of it later. I don't know why I remain curious about these contraptions, as I'm not planning to put it to any practical use myself. I have enough on my plate with other things I do, and when I think of the people I know who I believe are likely to be the most worthy opponents for it, my name is not on my list.Quote: mickeycrimmAl suggested this in depth New York Times article on the game.
http://www.nytimes.com/2013/09/08/magazine/poker-computer.html?pagewanted=1&_r=1
