Gorre18712 years, 1 month agoI always thought that the things stated in this video are true. A perfect GTO strategy certainly has and should have all betsizes on all boards. (The frequencies and distributions are obviously made in way to be unexploitable) The reason nobody is that far yet is that it is way way more tough to balance different betsizing for every spot than if u only have to balance it for one as you already said. But it is inevitably more tough for the person to adjust to that as you said in your video. There were some great posts by a Math PhD Student at 2+2 , link: http://forumserver.twoplustwo.com/showpost.php?p=28066755&postcount=62 So if u translate that into no limit that means that if u stick to one betsize for every spot all the time it is that same as playing a simpler version of NLHE and automatically the GTO version of that would loose to a more complex one where every betsize is avaible and will be used with perfect frequencies , distributions etc.
desperhate12 years, 1 month agothe great and only Phil is back on his best ! makes me feel happy to pay my fee on this site when I get such contents. Thanks! , this type of videos help me being more creative and thinkin deeper
James Obst12 years, 1 month agoAwesome vid Phil, I'm going to make a post in 'Other Games - Misc' that you might be interested in and I'd be very interested to hear your thoughts :).
Phil Galfond12 years, 1 month agoIt is applicable to PLO, however you need to be more careful since equities run so much closer. Still though, having more than one betsize prevents your opponent from being able to make good plans against you.
but what if i am not smart enough to split my range and keep all the permutations balanced :)
i hope future videos examine the entire process of building ranges in a spot where you plan to split your range(ie how to build 30% and 75% bet size range for the example you gave) i would really like to seem a detailed complete process so i can understand it all and apply the process to other situations. as a 6m guy one common spot where i am splitting my range(likely in unbalanced way) is OOP as the pre-flop raiser where i mix CB,XF,XC, and XR lines
I wouldn't worry so much about misapplying this concept. As I believe I said in the video, having these two betsizes randomized is superior to having just one betsize with your range, in my opinion (though you can add a few more bluffs to your existing range perhaps).
Balance is very important, but perceived balance is more important. Your opponents are not going to get a handle on your actual range, especially at 6max, until they've played 200k hands with you, IF they're observant. By then you'll probably have altered things anyways.
Jeff_W12 years, 1 month agoThis is one of the best strategy videos I've ever seen. Luckily most people are too lazy to do this kind of analysis! I agree that static bet sizing is a big leak vs Maximally Exploitative Strategy and top players only use it to avoid being even more exploitable if they used imbalanced dynamic bet sizing.
However, I'm not convinced by your $30 bluff example--villain is playing very exploitably if you he's vulnerable to this play and the counteradjustment seems simple--check all hands. You have weakened/capped your hand range by checking the flop and yet you're still committing very often to a turn bet. Of course, now we're getting into an equilibration exercise...
I didn't say we'd bet 100% of our air (though I realize now that I didn't say otherwise). Having our opponent check all of his hands to us would be a massive advantage (but we would obviously need to adjust for his new checking range, and then he can counter, etc). My point was simply to prove (in addition to giving you some excellent bluffing opportunities against an opponent if he doesn't adjust) that having multiple betsizes is strategically superior to having one bet size.
You're absolutely correct that this becomes an equilibration exercise as far as how often you bet, what sizings you use, and how you handle raises/calls, etc. But this is true of any spot in poker.
I am arguing that having multiple sizings gives you an inherent advantage against someone who only uses static betsizes in certain spots... so when both players reach their equilibrium frequencies (in theory), the player with multiple sizings will come out ahead.
Zachary Freeman12 years, 1 month agoAwesome video Phil.
It's intersting timing as well. If you have a minute take a look at a post I placed in Sean Leforts SBvSteals video. He has adopted a fold or 3bet single strategy for SB defense vs steals. I made an argument that although it takes some extra work to break up your continuing ranges into a mixed strategy of call fold and 3bet, the work is substantially worthwhile in profitability.
I have never thought of specifically building different balanced ranges for multiple bet sizes in such a organized way that you discussed. I think your methodology is superior to a single bet size strategy. It just takes more skill and work to build multiple ranges like that. And if built improperly could expose yourself to exploitations you can't as easily detect. Basically your game plan will be more complex which is good if you can govern it sufficiently but caution should be exercised to not loose a complete understanding of your ranges.
Phil Galfond12 years, 1 month agoHey Zachary. I'll take a look at that video thread once I finish responding to everyone here. Thank you. Any and all strategic contributions to the forums (and even the video threads) are appreciated greatly by me.
You're absolutely right that you can't lose understanding of your own range when you start splitting betsizes. That said, I think everyone is overreacting to how much work needs to go into a multiple betsize strategy before you can implement it. I started using one when I had done absolutely zero range analysis work.
Really enjoy these theory/analysis based videos. hope there is more to come
Here are some thoughts as I was watching the video:
1. when you are calculating the ev of betting air vs. an assumed betting range, howcome you did not include the ev of your value/protection bets? I'm just thinking that, if you bet non air hands and achieve an ev of x, maybe your ev could be higher if you include a portion of air in it too. Thus making your betting range complete, otherwise if you do not have air and have a disproportionate amount of value/protection with regards to your bet size, villain could technically exploit you by folding?
2. so after listening to your explanation of splitting up range with regards to bet size, would i be correct to say that the reason for this is to make villain make adjustments that he is not used to. because with each bet size, his optimal defending frequency changes and if he hasn't done the analysis he will make a mistake. BUT if he is aware of this strategy (somehow), his counter strategy would be to just fold less and add hands to both his cr and c/c range
3. a question that doesn't have to do with this video but for range construction. in a situation where your range is slightly inferior to your opponents (preflop equity), would you still be comfortable splitting up your range? or would a single play be considerable (something like never raise)?
1) I am not sure how I'd start to calculate the EV of my entire range and how the EV of my protection and value bets will be impacted. Obviously, that will be dependent on our opponent's adjustment to our multiple sizings.
The point of this video was simply to prove that having multiple betsizes is stronger strategically than static betsizes. I went through the range analysis for a couple of reasons:
-To show those who haven't done any analysis like that how to do it (roughly), and -To illustrate how villains checking range is not built to handle multiple sizings (meaning he needs to change his ranges, and without certainty of which betsize he will face, this will cause a lot of problems for him in his planning)
2) It wouldn't be correct to put it that way, I think. I'm not arguing that multiple sizings is the best vacuum play, but not better in the long run. I'm arguing that it's better in the long run, to the point that not having multiple sizings in NLHE in various spots is actually a clear leak in your game (in almost everyone's game currently).
3) Good Question. I think that there exist some spots in poker where you can get away with never raising (For example, when your opponent's range in PLO clearly consists of only the nut flush or the nut flush blocker). I would argue in most cases that having a very small raising range is better than none, but I'm not entirely confident that it's true.
If you could be more specific about an example of what you're asking, I can try to give a more specific answer.
Juan Copani12 years, 1 month agoGreat video !! Im really excited about how much i can work in my game looking this kind of things.
Like Mike said, i would love a video where you talk about the entire process of split a rangue so i can aplly myself to other situations. I´m doing this a lot when i am OOP beeing the preflop raiser too, but i´m not sure enough of how balance i am.
DirtyD12 years, 1 month agoThis is my favorite video on the site so far. I think it would be a good idea to make some videos in this series part of the Essential membership -- I would recommend that membership to people for this video alone. If you're only getting one video per month, I think a concept video like this has more value than a session review.
I'd be really interested to see someone with a strong grasp of theory make the case for why Phil is wrong.
I agree. I would love it if someone disagreed with me. Even if we find out that I'm wrong, we'll have advanced all of our knowledge in the process.
Phil Galfond12 years, 1 month agoMany people here want me to explain how I breakdown ranges and run the PPT sims and split my ranges up mathematically. I've never done that in my entire poker career. I share all of the knowledge I have and can get across to the RIO members. This video took me to calculations and range analysis that I'd never done before.
I will try to get more adept with PPT so that I can better teach you guys, but I am far from an expert in it. I've barely used it.
There are players who are much more mathematical than me (though I have always been strong in math), and who build their game using MUCH more study and analysis than I do (we hire guys like that too!). I've never done much studying in poker. I learn from videos, forums, friends, and mostly through practice and thinking.
The foundation of my game is logic, along with some psychology, which is what I try to share with you, as well as I can, in all of my videos.
I will work on studying a bit more, not for my own benefit, but for yours, so that hopefully down the road I'll be more qualified to offer strategy advice from that angle.
However, please understand that study and mathematical analysis has had next to nothing to do with me getting to the point that I am now in my career. I approach the game in a different way, and it's that approach that I think I'm most qualified to teach.
FX Brassard12 years, 1 month agoPhil, I love to approach poker the same way you do, mostly because I find that when a hand becomes a math problem everything becomes much less fun mostly. I also find, that even at the highest stakes NLHE 6 max games, you level yourself more often than not by obsessing about having your range balanced. Even in the toughest 3+ players games, the best players are playing a exploitative style and I think it will stay that way for a long time, because we are still far off from solving poker.
HOWEVER, with the trend being all this GTO stuff lately, arent you afraid the game will eventually pass you by? Especially since you play a lot of heads-up, where playing closer to GTO is most effective. I'm kind of afraid that one day everyone will have a close to perfect strategy down and I'll be completely lost against them because I haven't done enough homeworks in this area. Seeing a guy like you still having success against the best players out there kind of reassure me a little.
Phil Galfond12 years, 1 month agoIt's a legitimate concern, FX, that I may need to start doing that kind of work to keep up with the games.
I'm confident that I will put in the work necessary if I feel I start losing my edge, as I'm too competitive not to, but I honestly have no idea if/when that will happen.
DirtyD12 years, 1 month agoI'll venture another opinion, not really a counter-argument, but maybe a qualification. The video makes the argument that giving away information (in the form of a uniform sizing strategy) before your opponent acts probably isn't optimal. My understanding of the study of game theory optimal strategies is that it's assumed your opponent instantaneously adopts an optimal counter-strategy. If that's correct, the "surprise value" of springing a bet size that our opponent hasn't prepared for wouldn't be recognized in this model.
What having a mixed bet-sizing strategy DOES do, unquestionably, is introduce practical difficulties both for us and our opponents. If we've prepared properly, we should be able to handle those difficulties better than our opponents, because we've had the opportunity to study them away from the table, without time pressure and with the benefit of computer analysis tools.
I think the main reason most people haven't used mixed bet-sizing strategies, especially pre-flop, is the difficulty of implementing them over a large number of tables. "Okay, KTo...I 3x that 75%, 4x 15%, and 2x 10%...let me just consult my random number generator...Oh crap, I was looking at my CO chart, I'm actually in HJ...etc." That's a huge pain. In the past it was preferable to use simpler, easier to implement strategies over a large number of tables, because those strategies were more than adequate to defeat fish. We may be getting to a point in poker where it's necessary to introduce more difficulties into the game to get an edge on our opponents.
When top chess players play each other, they prepare their opening moves with the help of computers. Occasionally they find a move that's a clear improvement on existing theory, but that's rare. What they're really looking for is a sound line that hasn't been well explored and presents practical difficulties over the board. I think poker preparation for the best players might start looking more like that.
Phil Galfond12 years, 1 month agoI agree with most of your points, DD, but I want to be clear that the "surprise value" of a new betsize, while helpful, isn't the reason I think multiple betsizings is best.
With the OOP player having full information of our strategy, we are still better off having more than one betsize. Knowing he may be faced with either a full pot, 2x pot, .66x pot, or .3x pot bet before he checks the turn (even if he knows the frequencies/ranges of each) will be more of a disadvantage to him than knowing he will face either a .75pot bet or a check (and knowing the frequencies/ranges).
mike 12 years, 1 month agophil what do you think about using different CB sizes based on board texture? let say on the BUT v blinds for example - i often CB 1/2 pot on very dry boards and more on wet flops (when i choose to CB). last year i experimented with some very small CBs in this spot after watching a sixpeppers video. ie MR BUT and CB 1bb on very dry boards. i don't remember why i stopped but i do recall it creating some spew from regs who had no idea on how to adjust to it:)
do you think there could be something to the idea of using a much wider range of bet sizing on the flop? or does it become just too much fancy play and too many ranges to think about?
Phil Galfond12 years, 1 month agoI think altering cbet size based on board texture makes good sense (and I also believe in having more than one sizing on similar board textures as well).
I DO think having a much wider range of sizing is best, but I am not yet sure which sizings are ideal or at what frequencies. I do think you have to worry about just doing it for the sake of doing it, and getting too FPSy. Good points.
Sauce12312 years, 1 month agoHey Phil, check out Mathematics of Poker, page 154-157. They solve a [0-1] game, half street long, with one potsized bet remaining, and no checkraising allowed. The solution for the game has the in position player betting a different amount with every value betting hand and then balancing each betsizing with the appropriate number of bluffs. I'm not sure whether or not that result generalizes to some or any full scale poker game. At the very least, I'd say some full scale poker game situations are analogous to the toy game- situations where one player's range is heavily capped (for whatever reason).
It's been awhile, but if I remember, Chen and Ankenmann don't solve any no limit [0-1] games with checkraising. My experience has been that in no limit games with checkraising in position typically does not gain nearly as much EV (if any) from choosing different betsizes provided distributions are symmetric and there are no card removal effects, and distributions contain many different hand strengths (i.e. basically if the game is going to behave like [0-1] game). An example would be the flop in most NLHE situations.
Let's start with a simple case where I take sizing pot and 1/2pot. If opponent cannot checkraise, I bet pot with my stronger value threshold hands, and 1/2pot with my weaker hands. So, very roughly, I bet pot with my top 25% of hands, and bet 1/2pot with my next 12.5% of hands. Then I add 12.5% of bluffs to my potbetting range, and ~4.2% of bluffs to my half pot betting range. But if out of position can checkraise, and I do not alter my value thresholds, then OOP can exploit me by checkraising every stronger hand above the 1/2 pot range- i.e. he checkraises his top 25% along with 12.5% bluffs (I'll assume a pot sized checkraise). We'll say he also calls an additional 37.5% of the time for which I'll assume his expectation is 0. So, for IP: EV(check with [.25])=.75pot EV(bet with [.25])=.375(0)+.625(pot)=.625 Therefore, with a checkraise, IP must add some stronger hands to his value betting range for the smaller betsize in order to make it more profitable than checking, even though the hands in the small betting range were strong enough to bet when OOP could not checkraise. My conjecture (supported by some sims I've done) is that adding the balancing hands to IP's additional betsize in order to combat OOP's checkraise greatly reduces the EV gained from the multiple betsize(s) in situations where card removal isn't a significant factor, at least when both players have a distribution which is reasonably similar in strength to each others' and their distributions are linear (by linear I mean they have a reasonable amount of both strong, medium and weak hands).
I don't own Mathematics of Poker, but I understand what you're saying here. For clarification before my response:
In this toy game, is OOP checking with full range?
And, am I interpreting this correctly?...
"So, for IP: EV(check with [.25])=.75pot" - When IP player checks his top 25% of range, his EV with that range is .75*Pot
I'm confused about what situation this line represents. It looks like it's the EV of a bet with the top 25% of IPs range, but the numbers seem wrong to me, so I think I'm not reading it as you intended.
"EV(bet with [.25])=.375(0)+.625(pot)=.625" -
I still feel strongly that having multiple sizings in most spots is superior to not having them (which I don't think you're flat out disagreeing with), but I want to make sure I understand everything fully before responding.Sauce12312 years, 1 month agoPhil- Toy games are simple games which are intended to clarify some property of a more complex game. One of the species of toy games from MoP are called [0-1] games. In a [0-1] game, each player is randomly given a number between 0 and 1, where 0 is the nuts and 1 is the nut low.
The specific [0-1] game I'm referencing is heads up, and the OOP player checks his entire range (hence "1/2 street long" to use Chen and Ankenmann's terminology). OOP may only call or fold. IP's betting is no limit.
...
"And, am I interpreting this correctly?...
"So, for IP: EV(check with [.25])=.75pot" - When IP player checks his top 25% of range, his EV with that range is .75*Pot"
--- No, what I'm saying here is the EV of checking with the exact hand [.25] (i.e. the top 25th percentile hand) is .75 of the pot. That follows straightforwardly from the premises, since OOP always checks, therefore if IP checks OOP will hold a hand worse than [.25] 75% of the time, and a hand better than [.25] 25% of the time. So, IP's EV= .75(pot)+.25(0)=.75pot.
Phil Galfond12 years, 1 month agoThanks, Ben. I understand toy games, and have a rough understanding of the [0-1] game (the premise at least)... though clearly not enough to have considered that you obviously meant we're betting with .25 rather than top 25 :)
@Everyone: Ben and I talked a little on skype to clarify some things, but we both had to go. We'll make sure to keep the important parts of the conversation here for you to see and comment on.
Really enjoyed the video, takes me back the the theory vids you made on Bluefire that were pure gold and that's the reason I pay 100 bucks a month subscription.
I know lots of players are math based and they need there math junkie fix and that's understandable, but I much prefer it when you stick to what you do best.
Is there anyway you can get your old theory vids you made for bluefire to me? would love to re watch them and recap.
I think the other MTT instructors would do well to make the odd theory video on based on MTT's, seeing live sweat session, hand history's is educational to a point but so much more could be done imo.
Without seeming to suck balls, Jason Koon videos are amazing as his thought process is golden and learnt more from his vids than most.
Many thanks
Ben
huplz12 years, 1 month agoI find this situations when flop where x/x to be tricky to play on the turn oop and be balanced.
Mostly the ip range is going to be so weak that I feel like i want to bluff almost any turn and bet river often overbetting and that I kinda make up for that by leading with all my good hands also.
What i am saying tho is if we end up losing so much ev by doing a play that creats more balance is it might not be worth it?
My other questions is whats the definition of being explotibal here?
Lets say we lead 70% on the turn and when we check (30%) we fold 100% now can he start to check back and own us with delay cb making it so our turn bluff leads become more +ev.
what the ip player have to do then is to start bluff rasing turn bet when we check meaning he will start to put in more money in to the pot when his range is weaker then our.
If it is not we should start bluff XR him more on the flop.
Overall is it maybe best to just have a leading range on the turn and if not why wont it work and with what typ of hands / players?
tlnlloser12 years, 1 month agoHi Phil
Can You tell me , which Two Pair's You throw to (A) category and (B) category, and the same with flush draw No SD value and Qhi , Gutshots
Not much more conversation has gone on. I still feel confident about the video I produced.
The MOP example ignores the most important factor in the video (not that it wasn't supposed to)- Villain's turn checking/betting ranges, which as I pointed out, get very difficult to correctly plan once we introduce multiple sizings, since he now no longer has the information that he will face a check or a 3/4pot bet.
Amazing video. I noticed Jason Koon posted above so I thought that as a tournament player I'd add my thoughts. The strat discussions in the video are way above the level that's necessary for the midstakes tourneys I play, but I think the idea of multiple betsizings as part of what you might call a 'manipulative' postflop strategy is something that's familiar to a lot of tournament players, even though we might not define it as such.
I think the majority of tournament players are accustomed to situations where balance isn't quite so important as the perception of balance, since the same situation almost never comes up twice vs the same player - this means there are a lot of interesting things that we can do with betsizing to influence our opponents' ranges. Lately I've experimented with everything from cbetting 20% of pot, to overbet jamming turns for 3x pot, to minbet-leading rivers, and I'm beginning to feel more like these kinds of really extreme betsizings are an invaluable tool in tournament play - however, there are a lot of very ABC tournament regs who seem to write off anyone making plays like that as just being an exploitable button clicker who has no idea how to play postflop.
It's gratifying, therefore, to see that even one of the best NL thinkers in the world believes in thinking outside the box as a principle, and is considering things like betting 30% of pot on turn in the knowledge that there are only so many ways that the villain can respond to it. Thank you for helping me keep the faith that there's still room for a button clicker like me in a world of balance and GTO - I look forward to being able to move beyond donkaments one day and apply the strategy in this video at a deeper level.
Phil Galfond12 years, 1 month agoGlad I could give confidence to a fellow button-clicker :)Brian Rast12 years agoI feel I have personally seen many tournament players employ different bet-sizing. In fact, Phil Hellmuth does more pre-flop bet-sizing variation then anyone I have seen (ranging from limping to min-raising to 5x). I believe for the most part this is done in order to try to exploit opponents, rather than to create really balanced ranges... and obviously the more you balance ranges, the more that the exploitative advantages melt away... yet I am definitely persuaded by the idea that creating multiple ranges for different sizings could be more profitable than one sizing.
More "thinking out loud" videos please! Would be nice to have PLO examples on the next one!
OnceItRun12 years agoSick video! I always need a refresher on having to think more and not accept a play as best in any circumstance, which is the way to get better at poker. Reminds me of the so many spots that occur when someone just takes away the pot with a small bet w/o too much risk and I condemn him as playing bad, because in that exact spot with reverse roles I have a balanced-one-size range and I am "unexploitable".
Elio Fox12 years agovery interesting video Phil. I am not really sure about multiple betsizes but I can
neither argue against your logic nor against Ben's. I just wonder how we can
spin the thought further. Assuming having two betsizes is superior to having one,
does that mean three are superior to two? From how I understand it, your main
argument is complexity that comes along with multiple betsizes which causes our assumed
edge to increase. Following that logic, the more betsizes we have the morge edge
we should gain which of course doesn't work in practice if we want to be balanced.
Just curios to what you think about that.
Brian Rast12 years agoThis video is more or less why I joined Run It Once. Keep it up Phil.
By the way, what happened to your debate with Sulsky? You guys just got started... then it died. Really hoping you guys finish that debate... as much as I loved the video, I might have been enjoying the debate more (although hard to really say bc it was strangled as a child!)
Sauce12312 years agoWell, I think Phil and I agreed to disagree. We both agree on some foundational points, namely (1) that all strategic options have non negative value, (2) in some spots having multiple betsizes is definitely optimal and adds significant value to our strategy, and (3) multiple betsizings are often more exploitative than single betsizes.
Phil and I disagree a bit in our overall approach insofar as Phil generally does more reacting (because "exploiting" sounds kind of awful, e.g. "Phil's a great exploiter!", I'm going to say "reactive player" when I mean exploitative player from now on) than I do. That makes a lot of sense, since Phil is unbelievably good at getting inside of his opponents' heads and guessing what they'll do, much better at it than I am. So, for Phil, using multiple betsizings is probably a better idea than using them would be for me. And since all strategic options have non-negative strategic value, if we react better than our opponents, then adding additional betsizes will win.
I think where we disagree most meaningfully is that I think studying multiple betsizes is going to add little strategic value as compared to other areas of study- for example balancing our frequencies for one betsize. I tried to argue this point by showing that in games where players have relatively linear/similar distributions additional betsizes must be balanced or they'll have negative EV relative to a single betsize. In addition, as a practical matter, I think multiple betsizes are often very tricky to seize additional EV with, and fairly easy for opponents to exploit. I think Phil disagrees, and I can also chalk that up to him being a better reactive player than I am. Basically, I'm too big of a nit to try to guess right with two betsizes, I have enough trouble knowing what I have with one !
Brian Rast12 years agoI find your 3rd foundational point that you guys agreed on (according to Ben at least), that multiple betsizings are often more exploitative than a single one to be very convincing. I think the example I gave in a post above discussing PH speaks to this as well. I think based on this point alone, there is enough to look further into MBS's, as exploitative or reactive advantages alone can often times be large and worthy of study and/or implementation. And that is without bothering too much about the debate about being able to balance the multiple sizings, which is also interesting and sort of the eventual conclusion of including them for exploitative reasons (in fact with many things in the development of poker, an exploitative play that started working out strategically was implemented... then over time balanced to be part of an overall strategy - look at 3betting pre flop for example...)
When I was reading your debate above, fwiw, it seemed like you guys were in the midst of clarifications about the toy game that were happening before a debate was about to take place, and that actual debate about the subject never really got started. That is more or less where you stated that you think you guys disagreed most meaningfully - which also happens to be the main point Phil is trying to make (the strategic value of adding multiple betsizings), and really the crux of the matter imo.
I agree with you Ben, that balancing multiple bet sizings seems like a nightmare, especially "in game" when balancing one is already difficult enough. But, just because something is difficult, doesn't mean that it is unworthy of further study... or that it isn't actually the best strategic option. Yet, I would tend to agree that at this point, with what people seem to know and be able to do with poker in their study and how it manifests itself in their play, having multiple bet sizings is much more of an exploitative play, that will tend to make someone's game more unbalanced. Yet depending on the quantities of these differences, it could still be worth doing, even if you were not quite as balanced... and some of that goes in to a debate over "reacting" or exploiting your opponent, versus remaining balanced within your ranges... and to what extent each is important versus a particular opponent.
One of the primary reasons why multiple bet sizings seems appealing to me is that most people are still relatively unbalanced - and a strategic option that increases the amount that you can react and "re-exploit" opponents in the spots that they are unbalanced - is imo very worth looking in to. In a spot where an opponent is unbalanced, I feel strongly that a strategic option that exploits that unbalance at the cost of our own balance is well worth the trade off strategically. Thus, even if you are granted that balancing MBS's is a nightmare, and that it is perhaps too difficult to do in as balanced a way as 1 sizing, it could still be a valued strategic option, well worth studying, just for the times when our opponents are imbalanced in a spot, and we should consider MBS in that spot.
flomaster6512 years agoPhill great video. Golden info for highstakes regs! Me personally as midstakes 6max player like to see these kinds of vids, because they make me think more about the game. But I would really would like to see, is more exploitive based vids. This is what most of us need I think. I remember this vid on BFP where you talked about not exploiting 1 range, when somebody was raising too much. But exploit there calling range. Stuff like that increases my winrate instant. Im not saying I dont enjoy the advanced stuf. But I know there is a ton of exploitive stuff you do, which could make for some great content!!
T K12 years ago@ 18 minutes you made mistake in calculations ? EV when he calls. Should be ?
Calls (48.9%): -75 (-36,675)
Oliver Price12 years agoI've been trying to incorporate this to some degree in my 6-max game, both with pre-flop and post-flop sizings. This can get difficult to do properly when playing more than just a few tables, and even more so on turbo tables. But one advantage of this that I've found that hasn't been mentioned, is that as you're putting so much more thought into what you want your bets to achieve and good hands to balance with etc. it becomes so much harder to not slip into auto-pilot mode, just mashing the 3/4 pot button every street. This may not be relevant from a GTO/ theoretical viewpoint, but this is a problem that a lot of players have to deal with, and it seems to be a very positive by-product that employing these strategies can lead to. FWIW I also thought that the content was solid, and it gave me a lot to think about. TY for a great video.
Watching this video for the first time today has me seriously considering the upgrade, (despite it being hard to justify every month since I am playing only live poker and mostly for relatively low stakes) I am definitely going to think about this concept some more and then probably re-watch the video, but here is my initial reactions/questions.
1) As everyone seems to agree putting in the off the table work balancing multiple bet sizings seems somewhat nightmareish and could be just as hard to successfully implement in game over multiple tables over and over and over. I do not think that is cause to dismiss the strategy however and often times the hardest things to do can make the largest difference in you success.
2) One very relevant point that I have not seen mentioned, hoping I didn't miss it reading through the comments, is the face up nature of the strategy. Say we are playing a good thinking player in a heads up no limit match and we decide that against this opponent we decide that we should lower our Cbet frequencies by 15% because they are continuing too often. (I know that this is more of an exploitative adjustment as opposed to your more complete GT adjustment to our game, but it was just an example and people who play at the highest stakes against great competition can likely come up with a better complete strategy adjustment) My point is that lowering our Cbet frequency but keeping only one sizing is not going to be immediately apparent to our opponents, we are going to have an advantage on them until they catch on (assuming our adjustment was correct) where as splitting our range into 2 separate sizings is something that they will notice almost right away and certainly by the 10th time that different sizings come up ( so they know its not just a one hand thing). Do you think that making it so apparent to our opponent that there is something that we are doing that they need to adjust to is a big deal?
3) I could be way off on this one but my initial reaction is that there would have to be some serious analytics of our entire range in order to actually be able to tell if 1 sizing is better than 2, because if we are balancing both range as we obviously need to do there are going to be variables to consider: namely that we will some times be pricing in our opponents (with the small size) with hands that they otherwise would not have the odds to call with and we will not be getting as much value (again with the small size) with our valuebets against hands that would have called either bet. Obviously these are 2 simple examples of problems that would arise, and we can chose which hands to balance with but ultimately there is going to be some concessions made in order to be balanced (ie laying too good a price to call, or losing value with smaller sizing). Wouldn't you have to actually construct the entire range and then put numbers and EVs to all the different types of concessions and benefits of doing this in order to be able to tell whether is better or not? Or do you feel that the Pros of doing this outweigh the Cons by such a wide margin that a quantified analysis is not needed?
this is a very interesting video, and the discussion is awesome. I think I tend to agree with Sauce, while still finding a ton of value in what Phil is saying in the video. My feelings when I watched it was that Phil (while still attempting to approach his range construction in a balanced manner) was essentially advocating a change in bet sizing to exploit a knowledge gap from his opponent. IE when we adopt "standard" lines/bet sizings against good/thinking/attempting-to-be-balanced opponents we often allow them to apply a lot of considered and researched strategy to the situation. However when suddenly faced with a new strategy (with some semblance of balance behind it), in a time-limited game, a human opponent may not have the ability to correctly adapt to the changing strategy. This is essentially the core of all exploitative play. I change something about my strategy and, as long as it isn't obviously apparent what I'm doing, you may not adapt; or if you attempt to adapt you may make incorrect assumptions that lead to more profitable situations for me in the long run. This is the core of the both the micro and macro meta-game of any poker game, and it's interesting to hear great minds discuss the value of making such changes.
My guess is that the GTO/balanced based study of the game that Sauce and many top players advocate is almost certainly the most valuable way to spend your time studying poker once you've reached a fairly advanced understanding of the game. But I would also guess that any player not considering making small exploitative changes often in their game is leaving a lot of money on the table.
your penchant for thinking inspires many people, so congrats on that.
As for the video, I have one question/problem with it:
- Yes, you proved that with 2-bet-size strategy you make betting with air more profitable, however this is just one part of your whole range on the T. What about your profitability with the other parts of your range. Yes, you have increased your profitability with air, but you may have decreased your profitability with your value hands, or your semi-bluffs for example. And so, it may turn out that you have decreased the overall EV of the situation as a whole.
Hi, thanks for a great video! I want to make a small note. There is actually proof of some betsizes being best for ex. betting all-in with perfectly polarized range or betting x% of the pot on each street such that you bet all-in on the river (if possible or pot,pot,pot in pl deep case) in a static multi-street game allows for more bluffs and higher EV. But yeah, great thoughts here. While nlhe or plo haven't been solved yet you have a lot of room for actions.
I like how this gives me as a ''feel'' player a new way off making myself harder to play against a mathimatical or as it is called ''gto'' type off player. I never had any easy way off imcorporating math in my poker game. I have tried but for some reason it has made my game less good or made me think about the math more then making an actual play right. I have accepted the fact that because off this I will probably not become as good poker player as sombody who can calculate their ev to almost perfection in a certain situation but this will help me alot not to give as much money away as I could or even make me win against some ''gto'' players if the I can make myself harder to exploit because off this.
But what do you think as to vary your betsize almost every time? Sure maybe in some spots its bad to always to do it but in the big picture would it make me a harder player to play against if I wary it constantly. Like betting 2/3 of pot 2 times in a row then 1/3 1 time then 3/4 once then 2/3 again and then we go for pottbet and overbet aswell? In other way I am saying instead of balacing your range you balance your betsize as much as possible in different spot and in that way make yourself harder to play against matchimatically? As this decision is based more on a feeling/gameflow as combined with a strategy could it make me a better ''feel'' player?
Maybe I'm just misunderstanding the arguments here but it seems like this is more proof that just because your betting range is balanced on one half street doesn't mean that it is un-exploitable. Phil's example is just a case where the OOP player was exploiting the IP player in an a non-GTO way despite having a "balanced" leading range. The overall strategy was exploitable because the OOP player was removing too many of his value hands because he knew that the IP player would only use one bet size. i.e. he was able to meaningfully optimize for information he shouldn't have had.
So I think it's somewhat likely Ben and Phil are both right in different ways. Ben seemed to be arguing that if you assume that hand distributions are similar this is a probably a much smaller effect than the video implies, which is probably true. In this case the hand distributions aren't similar and the IP player can probably retain a higher percentage of his positional advantage by using a mixed strategy because it forces the OOP to play a more passive style. However, if the OOP player's overall turn strategy is optimal it probably doesn't really matter how many bet sizes the IP player uses. If anything I think the video makes the mixed strategy seem more offensive than it really is. It's really a defensive maneuver that recovers EV that the OOP player was stealing by optimizing for a single bet size.
Good post. I mostly agree with what you're saying.
It's really a defensive maneuver that recovers EV that the OOP player was stealing by optimizing for a single bet size.
The OOP player was stealing EV because of knowledge he shouldn't yet have (IP player's sizing when betting). Agreed completely, and it's an excellent way to put it.
However, if the OOP player's overall turn strategy is optimal it probably doesn't really matter how many bet sizes the IP player uses.
The word "optimal" is thrown around a lot in poker and is often misused (probably sometimes by me, since I don't have a GT background).
No one plays close to GTO, so are you talking about playing against a perfect player in theory? Or, do you mean you're playing against a player with a set strategy who won't deviate based on your IP strategy?
Either way, I don't think it matters for my argument.
If we are talking about in practice, surely every strong opponent can and should make adjustments based on your IP strategy. So, you'd agree then that in practice against a good player we should have a mixed strategy?
In theory, I'm not confident either way if mixing up your bet-size against a perfect GTO bot is beneficial or not, but I am confident that it can't hurt.
Further, if we want to talk theory... If we are attempting to build a strategy as close to GTO as possible, that means that we're trying to build a strategy that can't be exploited. Multiple turn sizings has to be part of an optimal strategy, otherwise it can be exploited by the extra knowledge OOP gets before acting. Agree or disagree?
Depending on your opponent, this strategy can either be used to prevent him from exploiting you or to punish his lack of balance or adaptability.
Call it offensive or defensive... it's simply the right way to play, and almost no one does it.
No one plays close to GTO, so are you talking about playing against a perfect player in theory? Or, do you mean you're playing against a player with a set strategy who won't deviate based on your IP strategy?
I mean playing against the perfect player in theory, which like you said isn't really important. I'm not even sure that what I said is correct and it would only be interesting against a non-adaptive bot playing perfect GTO poker anyway. I only really brought this up because I thought this might be part of this disconnect between you and Sulsky. It seems like he assumes that his opponents are going to play closer to GTO out of fear of exploitation so it's less valuable to him?
Further, if we want to talk theory... If we are attempting to build a strategy as close to GTO as possible, that means that we're trying to build a strategy that can't be exploited. Multiple turn sizings has to be part of an optimal strategy, otherwise it can be exploited by the extra knowledge OOP gets before acting. Agree or disagree?
Agree. Just to be clear, I don't think that means that you would have to have multiple bet sizes for every board/situation but you have to have them at least some of the time (probably most of the time).
In practice I'm not sure how you actually implement this. I imagine you'd use some sort of mixed strategy with the very top of your range and then split the better part of the rest of your value range? Maybe even put the better part of the rest in the smaller bet range since you'll have more bluffs and might need to protect against a re-raise, have less need for protection, etc? Probably depends on board texture too. I don't know... I could see people just using a smaller single bet size though just to avoid all the headache and negate it a little or just randomly using one strategy or the other with the same value range and adjusting their bluffing frequency.
Anyway, thanks for an interesting video. Makes me want to sign up and move to Canada...
Hi Phil, I think you can go both ways about building unexploitable strategy in some simplified one or two street games. You can optimize your sizing(s) for the whole range that you have based on prior actions or you can optimize you range (e.g. frequencies) for one betsize. In both cases it's possible to build unexploitable strategy but it's just harder to do for multiple betsizes so you don't see human players use more than 2 betsizes for each street. And if there's more than one Nash equllibrium for poker it might actually have something to do with choosing different betsizing schemes.
I liked the video, it made me think a lot, but I'm a little confused.
On the video, you show how the IP player can modify his betsizing to make his pure bluffs profitable, but that doesn't look too bad for the OOP player, for a couple of reasons:
-Even if the IP has a profitable bluff, that doesnt mean that the OOP is being exploited. As long as he has reasonable betting and check-defending frequencies the IP player cannot check back all his air to just make a small delayed cbet on the turn (that would be the real problem). In other words, if OOP bets 99% of the time and check folds 100% of the times when he checks (1% of the time), OOP player is not being exploited.
-The smaller the betsize the less bluffs you can include to your range if you want to remain more or less balanced. So your small blluffs won't occure very often.
i think it only makes sense to have multiple sizings. fkljg akdjfg lkdfgk is about how i feel right now after watching that video. it's going to be easier to click buttons, i will go from there :)
do you guys think sizings should be reflective of a players tendencies? for example, if we open the button and get called by the BB in two different scenarios, one in which the guy calls from BB 40% and the other who calls at 20% frequency, should this not impact our sizing on various boards, at least somewhat? I would naturally think we would not have to bet quite as large versus the guy with a huge call from bb frequency, am i wrong in thinking this?
Loading 70 Comments...
but what if i am not smart enough to split my range and keep all the permutations balanced :)
i hope future videos examine the entire process of building ranges in a spot where you plan to split your range(ie how to build 30% and 75% bet size range for the example you gave) i would really like to seem a detailed complete process so i can understand it all and apply the process to other situations. as a 6m guy one common spot where i am splitting my range(likely in unbalanced way) is OOP as the pre-flop raiser where i mix CB,XF,XC, and XR lines
I wouldn't worry so much about misapplying this concept. As I believe I said in the video, having these two betsizes randomized is superior to having just one betsize with your range, in my opinion (though you can add a few more bluffs to your existing range perhaps).
Balance is very important, but perceived balance is more important. Your opponents are not going to get a handle on your actual range, especially at 6max, until they've played 200k hands with you, IF they're observant. By then you'll probably have altered things anyways.
However, I'm not convinced by your $30 bluff example--villain is playing very exploitably if you he's vulnerable to this play and the counteradjustment seems simple--check all hands. You have weakened/capped your hand range by checking the flop and yet you're still committing very often to a turn bet. Of course, now we're getting into an equilibration exercise...
I didn't say we'd bet 100% of our air (though I realize now that I didn't say otherwise). Having our opponent check all of his hands to us would be a massive advantage (but we would obviously need to adjust for his new checking range, and then he can counter, etc). My point was simply to prove (in addition to giving you some excellent bluffing opportunities against an opponent if he doesn't adjust) that having multiple betsizes is strategically superior to having one bet size.
You're absolutely correct that this becomes an equilibration exercise as far as how often you bet, what sizings you use, and how you handle raises/calls, etc. But this is true of any spot in poker.
I am arguing that having multiple sizings gives you an inherent advantage against someone who only uses static betsizes in certain spots... so when both players reach their equilibrium frequencies (in theory), the player with multiple sizings will come out ahead.
It's intersting timing as well. If you have a minute take a look at a post I placed in Sean Leforts SBvSteals video. He has adopted a fold or 3bet single strategy for SB defense vs steals. I made an argument that although it takes some extra work to break up your continuing ranges into a mixed strategy of call fold and 3bet, the work is substantially worthwhile in profitability.
I have never thought of specifically building different balanced ranges for multiple bet sizes in such a organized way that you discussed. I think your methodology is superior to a single bet size strategy. It just takes more skill and work to build multiple ranges like that. And if built improperly could expose yourself to exploitations you can't as easily detect. Basically your game plan will be more complex which is good if you can govern it sufficiently but caution should be exercised to not loose a complete understanding of your ranges.
You're absolutely right that you can't lose understanding of your own range when you start splitting betsizes. That said, I think everyone is overreacting to how much work needs to go into a multiple betsize strategy before you can implement it. I started using one when I had done absolutely zero range analysis work.
Really enjoy these theory/analysis based videos. hope there is more to come
Here are some thoughts as I was watching the video:
1. when you are calculating the ev of betting air vs. an assumed betting range, howcome you did not include the ev of your value/protection bets? I'm just thinking that, if you bet non air hands and achieve an ev of x, maybe your ev could be higher if you include a portion of air in it too. Thus making your betting range complete, otherwise if you do not have air and have a disproportionate amount of value/protection with regards to your bet size, villain could technically exploit you by folding?
2. so after listening to your explanation of splitting up range with regards to bet size, would i be correct to say that the reason for this is to make villain make adjustments that he is not used to. because with each bet size, his optimal defending frequency changes and if he hasn't done the analysis he will make a mistake. BUT if he is aware of this strategy (somehow), his counter strategy would be to just fold less and add hands to both his cr and c/c range
3. a question that doesn't have to do with this video but for range construction. in a situation where your range is slightly inferior to your opponents (preflop equity), would you still be comfortable splitting up your range? or would a single play be considerable (something like never raise)?
thanks!
1) I am not sure how I'd start to calculate the EV of my entire range and how the EV of my protection and value bets will be impacted. Obviously, that will be dependent on our opponent's adjustment to our multiple sizings.
The point of this video was simply to prove that having multiple betsizes is stronger strategically than static betsizes. I went through the range analysis for a couple of reasons:
-To show those who haven't done any analysis like that how to do it (roughly), and
-To illustrate how villains checking range is not built to handle multiple sizings (meaning he needs to change his ranges, and without certainty of which betsize he will face, this will cause a lot of problems for him in his planning)
2) It wouldn't be correct to put it that way, I think. I'm not arguing that multiple sizings is the best vacuum play, but not better in the long run. I'm arguing that it's better in the long run, to the point that not having multiple sizings in NLHE in various spots is actually a clear leak in your game (in almost everyone's game currently).
3) Good Question. I think that there exist some spots in poker where you can get away with never raising (For example, when your opponent's range in PLO clearly consists of only the nut flush or the nut flush blocker). I would argue in most cases that having a very small raising range is better than none, but I'm not entirely confident that it's true.
If you could be more specific about an example of what you're asking, I can try to give a more specific answer.
Like Mike said, i would love a video where you talk about the entire process of split a rangue so i can aplly myself to other situations. I´m doing this a lot when i am OOP beeing the preflop raiser too, but i´m not sure enough of how balance i am.
I'd be really interested to see someone with a strong grasp of theory make the case for why Phil is wrong.
I agree. I would love it if someone disagreed with me. Even if we find out that I'm wrong, we'll have advanced all of our knowledge in the process.
I will try to get more adept with PPT so that I can better teach you guys, but I am far from an expert in it. I've barely used it.
There are players who are much more mathematical than me (though I have always been strong in math), and who build their game using MUCH more study and analysis than I do (we hire guys like that too!). I've never done much studying in poker. I learn from videos, forums, friends, and mostly through practice and thinking.
The foundation of my game is logic, along with some psychology, which is what I try to share with you, as well as I can, in all of my videos.
I will work on studying a bit more, not for my own benefit, but for yours, so that hopefully down the road I'll be more qualified to offer strategy advice from that angle.
However, please understand that study and mathematical analysis has had next to nothing to do with me getting to the point that I am now in my career. I approach the game in a different way, and it's that approach that I think I'm most qualified to teach.
HOWEVER, with the trend being all this GTO stuff lately, arent you afraid the game will eventually pass you by? Especially since you play a lot of heads-up, where playing closer to GTO is most effective. I'm kind of afraid that one day everyone will have a close to perfect strategy down and I'll be completely lost against them because I haven't done enough homeworks in this area. Seeing a guy like you still having success against the best players out there kind of reassure me a little.
I'm confident that I will put in the work necessary if I feel I start losing my edge, as I'm too competitive not to, but I honestly have no idea if/when that will happen.
What having a mixed bet-sizing strategy DOES do, unquestionably, is introduce practical difficulties both for us and our opponents. If we've prepared properly, we should be able to handle those difficulties better than our opponents, because we've had the opportunity to study them away from the table, without time pressure and with the benefit of computer analysis tools.
I think the main reason most people haven't used mixed bet-sizing strategies, especially pre-flop, is the difficulty of implementing them over a large number of tables. "Okay, KTo...I 3x that 75%, 4x 15%, and 2x 10%...let me just consult my random number generator...Oh crap, I was looking at my CO chart, I'm actually in HJ...etc." That's a huge pain. In the past it was preferable to use simpler, easier to implement strategies over a large number of tables, because those strategies were more than adequate to defeat fish. We may be getting to a point in poker where it's necessary to introduce more difficulties into the game to get an edge on our opponents.
When top chess players play each other, they prepare their opening moves with the help of computers. Occasionally they find a move that's a clear improvement on existing theory, but that's rare. What they're really looking for is a sound line that hasn't been well explored and presents practical difficulties over the board. I think poker preparation for the best players might start looking more like that.
With the OOP player having full information of our strategy, we are still better off having more than one betsize. Knowing he may be faced with either a full pot, 2x pot, .66x pot, or .3x pot bet before he checks the turn (even if he knows the frequencies/ranges of each) will be more of a disadvantage to him than knowing he will face either a .75pot bet or a check (and knowing the frequencies/ranges).
do you think there could be something to the idea of using a much wider range of bet sizing on the flop? or does it become just too much fancy play and too many ranges to think about?
I DO think having a much wider range of sizing is best, but I am not yet sure which sizings are ideal or at what frequencies. I do think you have to worry about just doing it for the sake of doing it, and getting too FPSy. Good points.
It's been awhile, but if I remember, Chen and Ankenmann don't solve any no limit [0-1] games with checkraising. My experience has been that in no limit games with checkraising in position typically does not gain nearly as much EV (if any) from choosing different betsizes provided distributions are symmetric and there are no card removal effects, and distributions contain many different hand strengths (i.e. basically if the game is going to behave like [0-1] game). An example would be the flop in most NLHE situations.
Let's start with a simple case where I take sizing pot and 1/2pot. If opponent cannot checkraise, I bet pot with my stronger value threshold hands, and 1/2pot with my weaker hands. So, very roughly, I bet pot with my top 25% of hands, and bet 1/2pot with my next 12.5% of hands. Then I add 12.5% of bluffs to my potbetting range, and ~4.2% of bluffs to my half pot betting range. But if out of position can checkraise, and I do not alter my value thresholds, then OOP can exploit me by checkraising every stronger hand above the 1/2 pot range- i.e. he checkraises his top 25% along with 12.5% bluffs (I'll assume a pot sized checkraise). We'll say he also calls an additional 37.5% of the time for which I'll assume his expectation is 0.
So, for IP: EV(check with [.25])=.75pot
EV(bet with [.25])=.375(0)+.625(pot)=.625
Therefore, with a checkraise, IP must add some stronger hands to his value betting range for the smaller betsize in order to make it more profitable than checking, even though the hands in the small betting range were strong enough to bet when OOP could not checkraise. My conjecture (supported by some sims I've done) is that adding the balancing hands to IP's additional betsize in order to combat OOP's checkraise greatly reduces the EV gained from the multiple betsize(s) in situations where card removal isn't a significant factor, at least when both players have a distribution which is reasonably similar in strength to each others' and their distributions are linear (by linear I mean they have a reasonable amount of both strong, medium and weak hands).
I don't own Mathematics of Poker, but I understand what you're saying here. For clarification before my response:
In this toy game, is OOP checking with full range?
And, am I interpreting this correctly?...
"So, for IP: EV(check with [.25])=.75pot" - When IP player checks his top 25% of range, his EV with that range is .75*Pot
I'm confused about what situation this line represents. It looks like it's the EV of a bet with the top 25% of IPs range, but the numbers seem wrong to me, so I think I'm not reading it as you intended.
"EV(bet with [.25])=.375(0)+.625(pot)=.625" -
I still feel strongly that having multiple sizings in most spots is superior to not having them (which I don't think you're flat out disagreeing with), but I want to make sure I understand everything fully before responding.
The specific [0-1] game I'm referencing is heads up, and the OOP player checks his entire range (hence "1/2 street long" to use Chen and Ankenmann's terminology). OOP may only call or fold. IP's betting is no limit.
...
"And, am I interpreting this correctly?...
"So, for IP: EV(check with [.25])=.75pot" - When IP player checks his top 25% of range, his EV with that range is .75*Pot"
--- No, what I'm saying here is the EV of checking with the exact hand [.25] (i.e. the top 25th percentile hand) is .75 of the pot. That follows straightforwardly from the premises, since OOP always checks, therefore if IP checks OOP will hold a hand worse than [.25] 75% of the time, and a hand better than [.25] 25% of the time. So, IP's EV= .75(pot)+.25(0)=.75pot.
@Everyone: Ben and I talked a little on skype to clarify some things, but we both had to go. We'll make sure to keep the important parts of the conversation here for you to see and comment on.
Really enjoyed the video, takes me back the the theory vids you made on Bluefire that were pure gold and that's the reason I pay 100 bucks a month subscription.
I know lots of players are math based and they need there math junkie fix and that's understandable, but I much prefer it when you stick to what you do best.
Is there anyway you can get your old theory vids you made for bluefire to me? would love to re watch them and recap.
I think the other MTT instructors would do well to make the odd theory video on based on MTT's, seeing live sweat session, hand history's is educational to a point but so much more could be done imo.
Without seeming to suck balls, Jason Koon videos are amazing as his thought process is golden and learnt more from his vids than most.
Many thanks
Ben
Mostly the ip range is going to be so weak that I feel like i want to bluff almost any turn and bet river often overbetting and that I kinda make up for that by leading with all my good hands also.
What i am saying tho is if we end up losing so much ev by doing a play that creats more balance is it might not be worth it?
My other questions is whats the definition of being explotibal here?
Lets say we lead 70% on the turn and when we check (30%) we fold 100% now can he start to check back and own us with delay cb making it so our turn bluff leads become more +ev.
what the ip player have to do then is to start bluff rasing turn bet when we check meaning he will start to put in more money in to the pot when his range is weaker then our.
If it is not we should start bluff XR him more on the flop.
Overall is it maybe best to just have a leading range on the turn and if not why wont it work and with what typ of hands / players?
Can You tell me , which Two Pair's You throw to (A) category and (B) category, and the same with flush draw No SD value and Qhi , Gutshots
phil did ben convince you that 2 bet size plan doesn't add EV?
ben i read and i think i understood MOP p154 but i am not sure how this is an argument for using only one bet size?
Not much more conversation has gone on. I still feel confident about the video I produced.
The MOP example ignores the most important factor in the video (not that it wasn't supposed to)- Villain's turn checking/betting ranges, which as I pointed out, get very difficult to correctly plan once we introduce multiple sizings, since he now no longer has the information that he will face a check or a 3/4pot bet.
Amazing video. I noticed Jason Koon posted above so I thought that as a tournament player I'd add my thoughts. The strat discussions in the video are way above the level that's necessary for the midstakes tourneys I play, but I think the idea of multiple betsizings as part of what you might call a 'manipulative' postflop strategy is something that's familiar to a lot of tournament players, even though we might not define it as such.
I think the majority of tournament players are accustomed to situations where balance isn't quite so important as the perception of balance, since the same situation almost never comes up twice vs the same player - this means there are a lot of interesting things that we can do with betsizing to influence our opponents' ranges. Lately I've experimented with everything from cbetting 20% of pot, to overbet jamming turns for 3x pot, to minbet-leading rivers, and I'm beginning to feel more like these kinds of really extreme betsizings are an invaluable tool in tournament play - however, there are a lot of very ABC tournament regs who seem to write off anyone making plays like that as just being an exploitable button clicker who has no idea how to play postflop.
It's gratifying, therefore, to see that even one of the best NL thinkers in the world believes in thinking outside the box as a principle, and is considering things like betting 30% of pot on turn in the knowledge that there are only so many ways that the villain can respond to it. Thank you for helping me keep the faith that there's still room for a button clicker like me in a world of balance and GTO - I look forward to being able to move beyond donkaments one day and apply the strategy in this video at a deeper level.
More "thinking out loud" videos please! Would be nice to have PLO examples on the next one!
neither argue against your logic nor against Ben's. I just wonder how we can
spin the thought further. Assuming having two betsizes is superior to having one,
does that mean three are superior to two? From how I understand it, your main
argument is complexity that comes along with multiple betsizes which causes our assumed
edge to increase. Following that logic, the more betsizes we have the morge edge
we should gain which of course doesn't work in practice if we want to be balanced.
Just curios to what you think about that.
By the way, what happened to your debate with Sulsky? You guys just got started... then it died. Really hoping you guys finish that debate... as much as I loved the video, I might have been enjoying the debate more (although hard to really say bc it was strangled as a child!)
Phil and I disagree a bit in our overall approach insofar as Phil generally does more reacting (because "exploiting" sounds kind of awful, e.g. "Phil's a great exploiter!", I'm going to say "reactive player" when I mean exploitative player from now on) than I do. That makes a lot of sense, since Phil is unbelievably good at getting inside of his opponents' heads and guessing what they'll do, much better at it than I am. So, for Phil, using multiple betsizings is probably a better idea than using them would be for me. And since all strategic options have non-negative strategic value, if we react better than our opponents, then adding additional betsizes will win.
I think where we disagree most meaningfully is that I think studying multiple betsizes is going to add little strategic value as compared to other areas of study- for example balancing our frequencies for one betsize. I tried to argue this point by showing that in games where players have relatively linear/similar distributions additional betsizes must be balanced or they'll have negative EV relative to a single betsize. In addition, as a practical matter, I think multiple betsizes are often very tricky to seize additional EV with, and fairly easy for opponents to exploit. I think Phil disagrees, and I can also chalk that up to him being a better reactive player than I am. Basically, I'm too big of a nit to try to guess right with two betsizes, I have enough trouble knowing what I have with one !
When I was reading your debate above, fwiw, it seemed like you guys were in the midst of clarifications about the toy game that were happening before a debate was about to take place, and that actual debate about the subject never really got started. That is more or less where you stated that you think you guys disagreed most meaningfully - which also happens to be the main point Phil is trying to make (the strategic value of adding multiple betsizings), and really the crux of the matter imo.
I agree with you Ben, that balancing multiple bet sizings seems like a nightmare, especially "in game" when balancing one is already difficult enough. But, just because something is difficult, doesn't mean that it is unworthy of further study... or that it isn't actually the best strategic option. Yet, I would tend to agree that at this point, with what people seem to know and be able to do with poker in their study and how it manifests itself in their play, having multiple bet sizings is much more of an exploitative play, that will tend to make someone's game more unbalanced. Yet depending on the quantities of these differences, it could still be worth doing, even if you were not quite as balanced... and some of that goes in to a debate over "reacting" or exploiting your opponent, versus remaining balanced within your ranges... and to what extent each is important versus a particular opponent.
One of the primary reasons why multiple bet sizings seems appealing to me is that most people are still relatively unbalanced - and a strategic option that increases the amount that you can react and "re-exploit" opponents in the spots that they are unbalanced - is imo very worth looking in to. In a spot where an opponent is unbalanced, I feel strongly that a strategic option that exploits that unbalance at the cost of our own balance is well worth the trade off strategically. Thus, even if you are granted that balancing MBS's is a nightmare, and that it is perhaps too difficult to do in as balanced a way as 1 sizing, it could still be a valued strategic option, well worth studying, just for the times when our opponents are imbalanced in a spot, and we should consider MBS in that spot.
Calls (48.9%): -75 (-36,675)
But one advantage of this that I've found that hasn't been mentioned, is that as you're putting so much more thought into what you want your bets to achieve and good hands to balance with etc. it becomes so much harder to not slip into auto-pilot mode, just mashing the 3/4 pot button every street. This may not be relevant from a GTO/ theoretical viewpoint, but this is a problem that a lot of players have to deal with, and it seems to be a very positive by-product that employing these strategies can lead to.
FWIW I also thought that the content was solid, and it gave me a lot to think about. TY for a great video.
Watching this video for the first time today has me seriously considering the upgrade, (despite it being hard to justify every month since I am playing only live poker and mostly for relatively low stakes) I am definitely going to think about this concept some more and then probably re-watch the video, but here is my initial reactions/questions.
1) As everyone seems to agree putting in the off the table work balancing multiple bet sizings seems somewhat nightmareish and could be just as hard to successfully implement in game over multiple tables over and over and over. I do not think that is cause to dismiss the strategy however and often times the hardest things to do can make the largest difference in you success.
2) One very relevant point that I have not seen mentioned, hoping I didn't miss it reading through the comments, is the face up nature of the strategy. Say we are playing a good thinking player in a heads up no limit match and we decide that against this opponent we decide that we should lower our Cbet frequencies by 15% because they are continuing too often. (I know that this is more of an exploitative adjustment as opposed to your more complete GT adjustment to our game, but it was just an example and people who play at the highest stakes against great competition can likely come up with a better complete strategy adjustment) My point is that lowering our Cbet frequency but keeping only one sizing is not going to be immediately apparent to our opponents, we are going to have an advantage on them until they catch on (assuming our adjustment was correct) where as splitting our range into 2 separate sizings is something that they will notice almost right away and certainly by the 10th time that different sizings come up ( so they know its not just a one hand thing). Do you think that making it so apparent to our opponent that there is something that we are doing that they need to adjust to is a big deal?
3) I could be way off on this one but my initial reaction is that there would have to be some serious analytics of our entire range in order to actually be able to tell if 1 sizing is better than 2, because if we are balancing both range as we obviously need to do there are going to be variables to consider: namely that we will some times be pricing in our opponents (with the small size) with hands that they otherwise would not have the odds to call with and we will not be getting as much value (again with the small size) with our valuebets against hands that would have called either bet. Obviously these are 2 simple examples of problems that would arise, and we can chose which hands to balance with but ultimately there is going to be some concessions made in order to be balanced (ie laying too good a price to call, or losing value with smaller sizing). Wouldn't you have to actually construct the entire range and then put numbers and EVs to all the different types of concessions and benefits of doing this in order to be able to tell whether is better or not? Or do you feel that the Pros of doing this outweigh the Cons by such a wide margin that a quantified analysis is not needed?
this is a very interesting video, and the discussion is awesome. I think I tend to agree with Sauce, while still finding a ton of value in what Phil is saying in the video. My feelings when I watched it was that Phil (while still attempting to approach his range construction in a balanced manner) was essentially advocating a change in bet sizing to exploit a knowledge gap from his opponent. IE when we adopt "standard" lines/bet sizings against good/thinking/attempting-to-be-balanced opponents we often allow them to apply a lot of considered and researched strategy to the situation. However when suddenly faced with a new strategy (with some semblance of balance behind it), in a time-limited game, a human opponent may not have the ability to correctly adapt to the changing strategy. This is essentially the core of all exploitative play. I change something about my strategy and, as long as it isn't obviously apparent what I'm doing, you may not adapt; or if you attempt to adapt you may make incorrect assumptions that lead to more profitable situations for me in the long run. This is the core of the both the micro and macro meta-game of any poker game, and it's interesting to hear great minds discuss the value of making such changes.
My guess is that the GTO/balanced based study of the game that Sauce and many top players advocate is almost certainly the most valuable way to spend your time studying poker once you've reached a fairly advanced understanding of the game. But I would also guess that any player not considering making small exploitative changes often in their game is leaving a lot of money on the table.
very cool stuff.
just wanted to chime in and say that i'm really glad that rio gave me a present on it's birthday instead of the other way around :)
and cool idea thought provoking vid as usual
Hi Phil,
your penchant for thinking inspires many people, so congrats on that.
As for the video, I have one question/problem with it:
- Yes, you proved that with 2-bet-size strategy you make betting with air more profitable, however this is just one part of your whole range on the T. What about your profitability with the other parts of your range. Yes, you have increased your profitability with air, but you may have decreased your profitability with your value hands, or your semi-bluffs for example. And so, it may turn out that you have decreased the overall EV of the situation as a whole.
this is such a good video, probably the best poker video ive ever seen (thus far) and now I'm really considering upgrading to Elite
Hi, thanks for a great video! I want to make a small note. There is actually proof of some betsizes being best for ex. betting all-in with perfectly polarized range or betting x% of the pot on each street such that you bet all-in on the river (if possible or pot,pot,pot in pl deep case) in a static multi-street game allows for more bluffs and higher EV. But yeah, great thoughts here. While nlhe or plo haven't been solved yet you have a lot of room for actions.
I like how this gives me as a ''feel'' player a new way off making myself harder to play against a mathimatical or as it is called ''gto'' type off player. I never had any easy way off imcorporating math in my poker game. I have tried but for some reason it has made my game less good or made me think about the math more then making an actual play right. I have accepted the fact that because off this I will probably not become as good poker player as sombody who can calculate their ev to almost perfection in a certain situation but this will help me alot not to give as much money away as I could or even make me win against some ''gto'' players if the I can make myself harder to exploit because off this.
But what do you think as to vary your betsize almost every time? Sure maybe in some spots its bad to always to do it but in the big picture would it make me a harder player to play against if I wary it constantly. Like betting 2/3 of pot 2 times in a row then 1/3 1 time then 3/4 once then 2/3 again and then we go for pottbet and overbet aswell? In other way I am saying instead of balacing your range you balance your betsize as much as possible in different spot and in that way make yourself harder to play against matchimatically? As this decision is based more on a feeling/gameflow as combined with a strategy could it make me a better ''feel'' player?
Maybe I'm just misunderstanding the arguments here but it seems like this is more proof that just because your betting range is balanced on one half street doesn't mean that it is un-exploitable. Phil's example is just a case where the OOP player was exploiting the IP player in an a non-GTO way despite having a "balanced" leading range. The overall strategy was exploitable because the OOP player was removing too many of his value hands because he knew that the IP player would only use one bet size. i.e. he was able to meaningfully optimize for information he shouldn't have had.
So I think it's somewhat likely Ben and Phil are both right in different ways. Ben seemed to be arguing that if you assume that hand distributions are similar this is a probably a much smaller effect than the video implies, which is probably true. In this case the hand distributions aren't similar and the IP player can probably retain a higher percentage of his positional advantage by using a mixed strategy because it forces the OOP to play a more passive style. However, if the OOP player's overall turn strategy is optimal it probably doesn't really matter how many bet sizes the IP player uses. If anything I think the video makes the mixed strategy seem more offensive than it really is. It's really a defensive maneuver that recovers EV that the OOP player was stealing by optimizing for a single bet size.
av,
Good post. I mostly agree with what you're saying.
The OOP player was stealing EV because of knowledge he shouldn't yet have (IP player's sizing when betting). Agreed completely, and it's an excellent way to put it.
The word "optimal" is thrown around a lot in poker and is often misused (probably sometimes by me, since I don't have a GT background).
No one plays close to GTO, so are you talking about playing against a perfect player in theory? Or, do you mean you're playing against a player with a set strategy who won't deviate based on your IP strategy?
Either way, I don't think it matters for my argument.
If we are talking about in practice, surely every strong opponent can and should make adjustments based on your IP strategy. So, you'd agree then that in practice against a good player we should have a mixed strategy?
In theory, I'm not confident either way if mixing up your bet-size against a perfect GTO bot is beneficial or not, but I am confident that it can't hurt.
Further, if we want to talk theory... If we are attempting to build a strategy as close to GTO as possible, that means that we're trying to build a strategy that can't be exploited. Multiple turn sizings has to be part of an optimal strategy, otherwise it can be exploited by the extra knowledge OOP gets before acting. Agree or disagree?
Depending on your opponent, this strategy can either be used to prevent him from exploiting you or to punish his lack of balance or adaptability.
Call it offensive or defensive... it's simply the right way to play, and almost no one does it.
Hi Phil,
I basically agree 100%.
I mean playing against the perfect player in theory, which like you said isn't really important. I'm not even sure that what I said is correct and it would only be interesting against a non-adaptive bot playing perfect GTO poker anyway. I only really brought this up because I thought this might be part of this disconnect between you and Sulsky. It seems like he assumes that his opponents are going to play closer to GTO out of fear of exploitation so it's less valuable to him?
Agree. Just to be clear, I don't think that means that you would have to have multiple bet sizes for every board/situation but you have to have them at least some of the time (probably most of the time).
In practice I'm not sure how you actually implement this. I imagine you'd use some sort of mixed strategy with the very top of your range and then split the better part of the rest of your value range? Maybe even put the better part of the rest in the smaller bet range since you'll have more bluffs and might need to protect against a re-raise, have less need for protection, etc? Probably depends on board texture too. I don't know... I could see people just using a smaller single bet size though just to avoid all the headache and negate it a little or just randomly using one strategy or the other with the same value range and adjusting their bluffing frequency.
Anyway, thanks for an interesting video. Makes me want to sign up and move to Canada...
Hi Phil, I think you can go both ways about building unexploitable strategy in some simplified one or two street games. You can optimize your sizing(s) for the whole range that you have based on prior actions or you can optimize you range (e.g. frequencies) for one betsize. In both cases it's possible to build unexploitable strategy but it's just harder to do for multiple betsizes so you don't see human players use more than 2 betsizes for each street. And if there's more than one Nash equllibrium for poker it might actually have something to do with choosing different betsizing schemes.
Greate video! Your videos are best! A lot to think about
I liked the video, it made me think a lot, but I'm a little confused.
On the video, you show how the IP player can modify his betsizing to make his pure bluffs profitable, but that doesn't look too bad for the OOP player, for a couple of reasons:
-Even if the IP has a profitable bluff, that doesnt mean that the OOP is being exploited. As long as he has reasonable betting and check-defending frequencies the IP player cannot check back all his air to just make a small delayed cbet on the turn (that would be the real problem). In other words, if OOP bets 99% of the time and check folds 100% of the times when he checks (1% of the time), OOP player is not being exploited.
-The smaller the betsize the less bluffs you can include to your range if you want to remain more or less balanced. So your small blluffs won't occure very often.
i think it only makes sense to have multiple sizings. fkljg akdjfg lkdfgk is about how i feel right now after watching that video. it's going to be easier to click buttons, i will go from there :)
do you guys think sizings should be reflective of a players tendencies? for example, if we open the button and get called by the BB in two different scenarios, one in which the guy calls from BB 40% and the other who calls at 20% frequency, should this not impact our sizing on various boards, at least somewhat? I would naturally think we would not have to bet quite as large versus the guy with a huge call from bb frequency, am i wrong in thinking this?
Phil was 3 years ahead in his thinking :D cool
Be the first to add a comment
You must upgrade your account to leave a comment.