wonder what kind of pair hands linus folds after calling flop, if he is calling down with this. it's a better bluffcatcher than any overpair due to the set blockage, but this has to be close to the very bottom of his made-hand range. i'd imagine TT and JJ are being folded somewhere, especially if they contain a diamond

TT-JJ??? the fuck. I think these guys playTT-JJ like solvers would, which are mixing 3bet/call on the flop, they will be 3betting the flop the most out of any overpair, for protection. and there is no way anyone folds TT-JJ on the river, that's insanely tight. folding a pair on turn should be virtually non-existent.

I'll see what the solver says!

bet/3bet TT/JJ on the flop? but why...? we want to just get it in with an overpair? that is something i never do... bet/3bet with overpairs. it's an interesting option but don't we always get it in kinda bad (either way behind or slightly ahead)? it also seems worse if stacks are deeper

no. being deeper has not too much of an effect, if you stop 3betting flops as much for value with marginal holdings for 130bb, 250bb, or 1000bb, then it allows them to c/r very aggressively as they always get to realize their equity. I'm not saying TT-JJ is marginal, it's a fine hand to get it in with for this stack depth.

lol hmm. i'll have to try it some time. just seems like ppl don't ever check/raise this flop with worse than TT for value, so it's always 2pair+ or some kind of semibluff. but yes we do get to deny equity from some of the weaker semibluffs

you don't have to. Most people won't take advantage of that exploit, so you won't have to 3bet light, deepstacked, unless your against one of those players. And i'm not so sure it applies as much when your IP. But it certainly applies a lot more oop.

anyway, i simmed it. It is actually very close with TT-JJ on river. A7 is slightly +ev call but very close, and is actually a better call than TT, for card removal. A9o does jam river for value some small amount. and they 3bet flop some small % of the time. it seem like linus would cbet this more than a solver would to me, though. But i havnt studied his hu game and dont know hu very well. well, it does actually make sense to make tight folds when its 1.5x overbet.

if the people you are playing are not c/r top pair at all here, then you can just hammer them with cbets. I guess if the people you are playing are like that then TT-JJ will be better off never 3betting

i've been playing zone (zoom) poker pretty much exclusively lately so i have basically no reads on any of my opponents. which i find kinda nice since i can just do my own thing (trying to play closest to GTO) and not think about player-specific exploitative plays very much.

that said i do think check/raising top pair here should be somewhat a part of GTO, since it has good blocker value, is a fairly strong hand, and needs protection. i don't see it often in my player pool though, probably because there are a lot of noobs there (its very low stakes). though i'm still kind of undecided about bet/3betting light for value; for the most part my bet/3bet range is extremely polarized, both IP and OOP. it's not a play i'm at all familiar with, which means i probably need to explore it a bit more

hmm would it be that bad to allow them to realize their light c/r equity? on the one hand they get to hit some cards and win some pots that they would otherwise had to fold or pay more to get to, but on the other hand we allow them to continue betting with worse hands and air and win more from them when they miss. unless of course we get bluffed at some point

trying to play close to gto is a theoretically unsound strategy at any stakes level, given the assumption that you are trying to maxmize your EV. There is a simple proof to this:

no one plays the same as solvers do.

since no one plays the same as a solver, it can never be the most +ev strategy to play like a solver yourself.

QED

3betting IP is not something that should be done often at all, and i would probably do it very rarely if those at your stakes arn't c/r top pair. You don't need to worry about this strategy. The solver cbets this flop around 52%, but the correct strategy at low stakes is to cbet around 90% or more, probably, as regs at low stakes under c/r and overfold to cbets.

You can't study peoples individual leaks in zone poker but you can study a populations and exploit that, and you can get away with highly exploitative poker since your anonymous, right? or is zone the non-anonymous one. I can't remember

stay away from runitonce videos imo, if you're watching them. they are dogmatic and there is a reason they pretty much all have single digit winrates

hmm but a solver's strategy (GTO) is still going to try to maximize EV, it just doesn't attempt any kind of exploitative play. because all exploitation is in itself exploitable, you just end up being exploited and it becomes a game of who can exploit better. whereas GTO will be optimal regardless of your opponent's strategy, so if they are making mistakes or trying to exploit, they should be punished for it naturally due to the interaction of the overall strategies

hehe it's kinda funny actually, my goal in poker as of the last few years or so has been to break even, sorta XD... i just want to understand the game at the basic level and play it with logical/mathematical excellence. poker in its final state is a break-even game anyway, as there are no advantages nor differentiating rules for any player, unlike many other games such as chess and starcraft. well, non-mirror-match starcraft anyway.

ah i haven't watched RIO in a long while now, but they do tend to approach the game in a purely mathematical way. i have to say, i do appreciate that kind of viewpoint... i feel like that's where a kind of 'scientific' progress is made with regards to the game. but it is definitely easy to get caught up in the dogma and forget about other possible ways to play that may be more 'optimal' or advanced. also really curious about how solvers find solutions (have never really used one before)

edit: yep everyone in zone is anonymous

On January 14 2018 11:23 fira wrote: because all exploitation is in itself exploitable, you just end up being exploited and it becomes a game of who can exploit better.

no it doesn't, I have significant empirical evidence that suggests otherwise. but think about it, people have to know what is optimal to know they are being exploited...this is why you can get away with it for a very long time, even against the best players in the world. I find it hard to believe that you wont get away with it for years in anonymous games, since the meta itself has to change there.

On January 14 2018 11:23 fira wrote: i just want to understand the game at the basic level and play it with logical/mathematical excellence. poker in its final state is a break-even game anyway, as there are no advantages nor differentiating rules for any player, unlike many other games such as chess and starcraft. well, non-mirror-match starcraft anyway. ah i haven't watched RIO in a long while now, but they do tend to approach the game in a purely mathematical way. i have to say, i do appreciate that kind of viewpoint... i feel like that's where a kind of 'scientific' progress is made with regards to the game.

playing exploitatively is more mathematically rigourous as attempting to play close to gto. Imo playing close to gto is a pseudo scientific approach to poker. These libratus/piosolver type machines have only some mathematics, only have machine learning algorithms that learn through trial and error. They have, for example, no software that allows them to make use of Bayesian inferences, no software for finding patterns in human timings for each decision. no mathematical modelling for common sense observations in poker, no mathematical models based off human psychology, no ability to look at decision tree frequencies of human players and understand them. no understanding of basic pot odds... If you look at solvers, their use of mathematics is a very narrow range compared to what a human can do. It is pseudo scientific because it doesn't try to maximize EV against a human, by ignoring all the ways you can theorize about how humans play poker.

I think that right now if you stuck a libratus equivilent ability bot for 6max into midstakes or lower, there would be human players that would outperform it in long term winrate.

i don't doubt that humans can still outperform robots in poker. however there is a scientific approach to the game with factual conclusions that cannot be denied. things like using relevant blockers to strong holdings, appropriate bet sizings depending on board texture and previous player action (regardless of individual player considerations), considering what hands to bluff with and bluffcatch with. things like this are objective pillars in the game that, at the highest levels of the game, simply cannot be ignored, because they are based on the rules of the game itself and the mathematical implications of such rules. whereas there are a lot of exploitative considerations that aren't actually very meaningful in the long run; things like timing tells, where a player could employ "false tells" and such in order to make such observations meaningless. the exploiter could easily get exploited in return.

one interpretation of exploitative play that i do like is where the exploiter is essentially capitalizing on the mistakes of their opponents, in the sense that the exploiter isn't necessarily "trying" to exploit their opponent directly, but rather playing in a way that allows the opponent's mistakes to surface wherein they can be seen as mistakes (sub-optimal play, and there are definitely objectively bad plays, most obvious ones being checking back the river with the nuts, calling allin with the nut lo, etc). i feel that any kind of direct attempt to exploit an opponent's habitual mistakes is only setting oneself up to be exploited in return. it can be very profitable in the short run, but at the game's highest levels (arguably nobody is playing at such a level, but still) such mistakes should not be made ever, and so the exploitation of such mistakes become irrelevant and probably counterproductive

of course card removal is important, i never said it wasn't. that is factored into an exploitative game style just as much as any other style. However, if your trying to decide whether to river barrel someone, if they overfold by 5%, then the math tells you to play a highly exploitable bluff frequency, and to stop caring about card removal.



your trying to make your approach sound more scientific and mathematical, however a strategy of trying to play like a gto bot, it's based on false psychological assumptions. Which is that regs will adjust so well that you shouldn't bother with an exploitative strategy. Linuslove and ohheycindy managed to exploit nosebleed regs for a several months before everyone copied them, that was against worlds best, adaptive players. Those regs actually had to do serious solver analysis to figure out how they were being exploited. at lower stakes you can get away with very obvious exploits for a very long time, i wish i could show you the evidence, but it would give too much away. It is surely, very wrong though.

so in summary of what i've been saying, the 'trying to play gto' strategy is based on bad mathematics and false tacit assumptions of psychology.

i'm just saying that there are elements in poker that are eternally important and relevant, and elements that aren't. card removal is one of those important elements because in 5, 10, 100, 5000 years, it will be exactly the same. however, things like meta-considerations and player-specific plays are always changing depending on who you are playing against, so it's much more frivolous to discuss and utilize. i think bet sizing depending on board texture is another very important and static element of poker, along with more obvious things like pot odds, implied odds, etc.

it's interesting to see how the general strategy of bet sizing has evolved over the years... nowadays you see the 1/4 to 1/3 cbet size on very many board textures (possibly all boards) but before it was always standard to do something like 1/2 to 2/3. part of that may be the exploitative adaptations players are making against one another, and is just a current trend, but i do believe these trends have a more objective message behind them, rather than simply playing a rock to an opponent's scissor. for example i don't think a very large cbet on an extremely dry board is ever a good option at the highest level of play, simply because bluffing hands don't have enough equity to be putting in so much money being so behind, and it makes it rather easy for opponents to play their hands.

anyway i think a lot of the 'psychology' behind poker is generally kinda trendy stuff that won't last in the game's final stages. sure it is a human vs human game, but at the highest level it's really all math imo. if you are playing in a way that can be exploited, then you are making game theory mistakes. also i don't think cindy's play is good at all... i've seen a lot of their hands and most of them i find kinda to very bad. linuslove is extremely, extremely good though. like, the two are incomparable imo

also if regs had to do solver analysis to figure out how they were being exploited, it has to say something about solver analysis being something of high relevance to the game's objective strategies in maximizing EV. that's one thing i don't really understand about your general approach - you seem to place a lot of emphasis on solver strategies, yet at the same time you also seem to be saying that it doesn't really matter much

However, if your trying to decide whether to river barrel someone, if they overfold by 5%, then the math tells you to play a highly exploitable bluff frequency, and to stop caring about card removal.

while this may be true, at the highest level, players should not be overfolding at all, so such a thing becomes irrelevant. if they are overfolding then they are playing badly, which will hurt them on its own. likewise, if you decide to overbluff rivers against someone who is in fact not overfolding, then your play will end up hurting you on its own. exploitation ends up hurting itself unless done against 'weaker' opposition

you keep saying about how people can adjust, but there are two spots in this post which show you could be exploited, and wouldn't know what the proper adjustment is:

1) you didn't know flop 3betting marginal value hands deep stacked is part of a game theory optimal strategy to prevent people from realizing equity with all their flop raises.
2) you think betting large on dry boards can never be good. This is simply wrong, game theory optimal strategies will cbet overbet size on 224, 225, 223, 442, ecct rainbow boards, with middle pocket pairs and suited connectors at a high frequency, in 3bet pots in BB vs BU and SB vs BU positions

since you don't know these trivial gto strategies, which are learned by someone who's studied solvers even a tiny amount, you can be exploited. So lets say i decide on trying to play close to 'gto', how are you going to just adjust vs my overbets on dry boards, and high frequency flop raising when deepstacked? You won't adjust properly unless you actually learn what game theory optimal play is. You may also adjust but in a way that makes you worse, if you don't understand the optimal strategy.

you arn't making 'game theory' mistakes by playing exploitatively, that sentence doesn't even make sense. I'm pretty sure a game theory professor is going to suggest you go paper every time if someone else goes rock every time, and not say it's a mistake to do so.

of course ohheycindy is a good player, he beats 25/50+ on stars. you may think he is a bad player but your just wrong.

I never said solver strategies arn't important. They are extremely important, You need to study with solvers a ton to play a good exploitative strategy.

i'm not arguing for my own skill in the game, there is obviously plenty of stuff i don't know about at all. i'm just arguing for my approach to the game, which is one of non-exploitation. like i said, successful exploitation implies that the one who is being exploited is exploitable. being exploitable by definition (imo) means you are making mistakes. therefore being exploitable is not optimal. therefore exploitation has no place in optimal play.

are you saying that being exploitable is naturally occurring in optimal play?

consider rock paper scissors. the solved strategy is to play each rock, paper, and scissors a third of the time. such a strategy is unexploitable by all other possible strategies. this is the kind of strategy that i'm aiming for in poker. but it's a bit different because in rock paper scissors, no option is a objectively a mistake. which in poker is not true, as there are objectively bad plays in poker.

"trivial gto strategies" what makes a strategy 'gto' by your definition? i was just stating some of my thoughts, i could definitely be wrong but, betting large on a dry flop was just an example of a play that i think is objectively bad. i might be wrong, but that's not the point. the point is that there *are* plays that are objectively bad. such as folding 100% of hands preflop. or going allin with 100% of hands preflop.

you think betting large on dry boards can never be good. This is simply wrong, game theory optimal strategies will cbet overbet size on 224, 225, 223, 442, ecct rainbow boards, with middle pocket pairs and suited connectors at a high frequency, in 3bet pots in BB vs BU and SB vs BU positions

here again you are confusing me; why even mention "game theory optimal" if you don't think GTO is important? and what makes such strategies GTO in the first place? i don't understand, it's like you are arguing for GTO and against it at the same time ??

well, from what i've seen of cindy, i dislike most of his plays. i can certainly be wrong, but that's kind of off topic

well my biggest problem with exploitative play is still, it seems to be targeting another player's mistakes. overfolding, underfolding, overbluffing, underbluffing, etc. what about a player who is not making any mistakes? how would exploitative play have a place against such a player? (yes i realize such kinds of players probably don't exist yet, but again i'm trying to think about what solved poker looks like, not what is most profitable at this current day and age)

(btw i wasn't pointing to boards like 224 as dry boards. i was thinking more like K72 rainbow. on 224 it's not particularly dry because any 2 cards have a 6 outer to top pair. and i do think it's viable to bet larger on such boards. but i still think a large cbet on a board like K72r would be bad. the majority of hands on K72r would have no pair no draw and be close to drawing dead against just top pair. which is why i think it's bad to bet large; u'd have to be betting a lot of zero equity hands with the same sizing and lose a lot of $ doing so and getting called by top pair. anyway this is just a random arbitrary situation, not the main point of what i'm saying)

re-read the above. you are misunderstanding all my points but i think they are clearly written.

can you list your points again? i don't think i am misunderstanding your points, i just don't see what you are saying. there is not enough understanding for there to be any misunderstanding o.o

well, one possibility of what you are saying is that, theoretically optimal play involves an element of exploiting and being exploitable. if that is what you are saying then it would at least be a considerable claim. either way you'd have explain a bit more what you mean

and i feel i have communicated my points very clearly too, especially with my most recent previous post... would be helpful if you comment a bit on what i said in that post... i asked a lot of questions there!

On January 15 2018 12:35 fira wrote: i'm just arguing for my approach to the game, which is one of non-exploitation. like i said, successful exploitation implies that the one who is being exploited is exploitable. being exploitable by definition (imo) means you are making mistakes. therefore being exploitable is not optimal. therefore exploitation has no place in optimal play.

you must mean game-theory-optimal when you say optimal. Some people mean the most +ev strategy when they say it's the optimal strategy.

On January 15 2018 12:35 fira wrote: Show nested quote + you think betting large on dry boards can never be good. This is simply wrong, game theory optimal strategies will cbet overbet size on 224, 225, 223, 442, ecct rainbow boards, with middle pocket pairs and suited connectors at a high frequency, in 3bet pots in BB vs BU and SB vs BU positions here again you are confusing me; why even mention "game theory optimal" if you don't think GTO is important? and what makes such strategies GTO in the first place? i don't understand, it's like you are arguing for GTO and against it at the same time ??

I was making a point. You assumed that people can just make adjustments. I used two examples where you showed a lack of knowledge of game theory optimal strategy in poker, and would not be able to make an adjustment, since you didn't know what gto was. and you kinda need to know to counteract. I'm basing this off my study of solvers which are close to what gto would be.

any exploitative player will need to know what optimal poker is, so they can see how their opponents deviate from it. This will require many hours of studying simulations. That's why i mention gto poker.

On January 15 2018 12:35 fira wrote: well my biggest problem with exploitative play is still, it seems to be targeting another player's mistakes. overfolding, underfolding, overbluffing, underbluffing, etc. what about a player who is not making any mistakes? how would exploitative play have a place against such a player? (yes i realize such kinds of players probably don't exist yet, but again i'm trying to think about what solved poker looks like, not what is most profitable at this current day and age)

um ok, that's just weird. Why would you care about non-existent players that make zero deviations from gto? Clearly exploitative play has no advantage against someone like that. But that doesn't exist and i doubt it will ever exist.

you must mean game-theory-optimal when you say optimal. Some people mean the most +ev strategy when they say it's the optimal strategy.

yes i meant GTO. the most +EV strategy would depend on how your opponent is playing, and that would not be GTO (unless they are playing GTO themselves)

I was making a point. You assumed that people can just make adjustments. I used two examples where you showed a lack of knowledge of game theory optimal strategy in poker, and would not be able to make an adjustment, since you didn't know what gto was.

well, i'm not really talking about how people generally play. there are all kinds of players, and my approach has been to completely ignore how someone plays, and simply play as close to GTO as possible, or at least study it best that i can so i can understand GTO as much as possible.

you dismissed this kind of approach as pseudo-scientific though, which i have to disagree with. it is very much scientific because it is based in undeniable facts like card removal, board texture, SPR, etc, rather than assumptions about how an opponent may be playing.

also, the way you talk about GTO, you seem to agree with me that GTO exists, and that it is the solved state of the game. and you do seem to respect that. you also seem to value studying GTO. we both seem to agree that GTO is the game's ultimate strategy. so i don't understand why you think trying to play GTO is not a good approach.

any exploitative player will need to know what optimal poker is, so they can see how their opponents deviate from it. This will require many hours of studying simulations. That's why i mention gto poker.

hmm i don't think all exploitative plays need to know about optimal poker. some exploitative plays have nothing to do with theory whatsoever. for example in live play if you have extremely keen senses and can know what your opponent has by just looking at their expression and posture, you can exploit them very hard that way. but you don't need to know a thing about poker beyond the very basics to do that. same thing with stuff like timing tells, and mostly everything related to psychology

um ok, that's just weird. Why would you care about non-existent players that make zero deviations from gto? Clearly exploitative play has no advantage against someone like that. But that doesn't exist and i doubt it will ever exist.

because i'm trying to understand as best as i can what GTO looks like. understanding of GTO may help an exploitative player, but understanding of exploitative play doesn't always help in understanding GTO, as some of it has very little to do with game theory

On January 16 2018 02:27 fira wrote: you dismissed this kind of approach as pseudo-scientific though, which i have to disagree with. it is very much scientific because it is based in undeniable facts like card removal, board texture, SPR, etc, rather than assumptions about how an opponent may be playing.

well, it's pseudo scientific if the assumption is your trying to maxmize ev. If your trying to not maximize ev, and be unexploitable then sure it's scientific approach. whether its based on things like card removal is irrelevant, since both approaches use that information to the fullest amount

On January 16 2018 04:19 Stroggoz wrote: Show nested quote + On January 16 2018 02:27 fira wrote: you dismissed this kind of approach as pseudo-scientific though, which i have to disagree with. it is very much scientific because it is based in undeniable facts like card removal, board texture, SPR, etc, rather than assumptions about how an opponent may be playing. well, it's pseudo scientific if the assumption is your trying to maxmize ev. If your trying to not maximize ev, and be unexploitable then sure it's scientific approach. whether its based on things like card removal is irrelevant, since both approaches use that information to the fullest amount

hmm the assumption is to first be unexploitable, and then to maximize ev. there are probably some strategies that are unexploitable, but do less well than others in maximizing ev. but yes, being unexploitable comes before maximizing ev, otherwise it would be exploitable and that would not be GTO.

lol baal XD

Libratus was adjusting on the go afaik btw

Oh and Fira a player that plays optimally will literally never exist, not even close to it , people aren't very close even with pio out for years with locked bet sizings, but when you have no fixed sizings and can mix combos at will the human brain is far from having the capability of figuring out the mixes etc.

Libratus has shown that in HU

And you always have poker theory even live with reads lul, none of that matters really there's still optimal strategy but you also add live tells and timings to it

When you'd theoretically know the gto strategy, you know how far he deviates from it and you can then much more precisely deviate the other way than when you're just guessing

On January 19 2018 09:46 LemOn[5thF] wrote: Oh and Fira a player that plays optimally will literally never exist, not even close to it , people aren't very close even with pio out for years with locked bet sizings, but when you have no fixed sizings and can mix combos at will the human brain is far from having the capability of figuring out the mixes etc.

if that is true then i guess poker is ultimately a game for robots :/

On January 19 2018 09:52 LemOn[5thF] wrote: And you always have poker theory even live with reads lul, none of that matters really there's still optimal strategy but you also add live tells and timings to it When you'd theoretically know the gto strategy, you know how far he deviates from it and you can then much more precisely deviate the other way than when you're just guessing

well that's the thing, you have to understand general GTO strategy, which is not a simple task. and knowing how to exploit doesn't really help in your understanding of GTO for the most part

No lol
you're looking at it all wrong
That's great news that a bot crushes HU and eventually will destroy 6max and people can't even come close to it
As long as there is control vs AI/Bots online it just means there's always edge in specific spots to be had, and no unbeatable human player will exist you'll always be able to pick your spots in situations you've worked out better

Remember DC before it exploded?
Blah123 and Improva, the two clowns that were saying poker's over, it's solved
Were all mysterious
And turns out both were clueless idiots that were gone in months and sunk DC
And that was like 2011 lol

go back to the Libratus challenge
It was doing some crazy stupid betsizing with fractions of combos
Showed a short glimpse of what real powerful optimal strategy is in NL
Fucking confusing xD

eh how am i looking at it wrong?

hmm i'm not saying that poker is solved. i don't know enough to say that (but i dont think it is... it may never be solved because stack depth can approach infinity o.o). and i dunno if there even is a bot that can crush HU. does libratus do that? how would libratus fair against the best human players of today, i wonder. and if it really does crush, why doesn't the programmer and friends just use it in the highest stakes online and destroy all the nosebleed regs? oh right they're already doing so haha xP

hmm yeah i am really quite interested in all the crazy sizings that are possible in NL, deep. its really interesting to try and understand how such strategies do so well compared to more "standard" lines. anyway these are just more reasons to study and play GTO, even if we lowly humans could never actually reach the 'perfect' state of poker-playing. we can try to get as close as we can, but more importantly we can understand why GTO is the way it is.

i downloaded the free version of piosolver and watched a demo tutorial on youtube. it's interesting but i'm not sure how the program comes up with the numbers and optimal plays and everything? is it playing against itself for a massive number of hands and doing trial/error to find the most +ev decisions based on all possible opponent decisions? something like that? seems like that would be a LOT of simulating. also since you have to enter the betsizing yourself, it can only run the simulation for that one sizing... so you'd have to run it for each sizing to find the optimal sizing? idk i am le noob~

Just Google it, were you gone when it was going on? :D
It already played and beat topHU regs.

Why not use it in games ? It costs shitloads just to run the fucker
It was a supercomputer built and operated by research teams not just some neckbeard's programme .


Also Pio doesn't use neural net whatever to play itself, that's PokerSnowie , pio literally calculates all possibilities aka the full tree