kingkong

I think the question is mostly how often we run into Q9. We are still vs 2 players. I guess how often SB has Q9 depends the open size. If we want to minimize the proportion of time we are called by Q9, we should bet small. But it is true we have the polar range (at AQ), but OOP is actually capped at Q9. But I think also considering how wide the BTN open, a lot of hands might want to bet more merged. I don't really know.. that's why I tought this spot is difficult..

I think if BTN don't really want to bet AJ QJ, then the argument for betting merged just would not apply.

Also, I think of it, betting also have equity advantage since we'd fold a lot Ax 9x, and other things depend on the size... But yes I think we should bet mostly under 1/3 and probably go more polar on the turn and have an easy check back (still on the turn) with QJ AJ and many Kx and give ups, and... I also don't really know how to play the turn to be honest...

What to you think of AJ and QJ, would you bet it small sometimes ? I think that might be the reason to go small on the flop and size up on the turn. I don't really know. The tool I use for this spot is snowie but I have mix confidence in it.

But really in your example AA vs 76hh both players have a ''value hand'' (of course it's not clear cut, but 76hh has better value than pot odds). I think you fold about MDF to give bluffs an incentives. Still your example is very interesting. Because even if you increase the bet size, 76hh will lose value but will never become a 'bluffcatcher'. It could however become a 0EV call. So it seems like the concept of 'MDF' don't really apply vs a range of only draws, since it's only a pot odds question.

But this makes sense with what I said before, that you would defend roughly more** than MDF vs a raise once you bet. If we're on the flop deciding bet or check, there are many reasons why MDF should not apply.

For example, if you force AA to check, then 76hh will never bet. Ok obviously it has a better EV check.

But if you would bet it, AA will check-raise you enough to greatly reduce the EV of a call. You might not be able to call. That's not the reason why you don't bet, but there is a correlation.

Well if you don't defend MDF vs a raise, your opponent can only raise instead of folding. So you would never bluff, so then they start folding again. So that's actually what is going on. When you fold too much vs their raise, you actually have too many ''bluffs'', so you have to remove some.

Ok, I understand the defense frequency depends on many factors. But when it's vs. a raise, from my understanding, it should not be less than the typical MDF. Otherwise you should not bet at all, bc your opponent should never fold at all, because if they raised, you would fold too much.

If however your range would be somewhat weaker coming on the flop, it can be good to check-fold more than the typical MDF. Or it might even be good to have a betting range and overfold after you check compare to MDF.

But if for some reason you take a decision to have a betting range, you should be able to defend more than MDF vs a raise.

I'll solve it tomorrow morning. I'm already running something in PIO.

Yes I'm pretty sure it solve one bet at a time. Actually Snowie says they can use 1/4, 1/2, pot or 2x pot. But what I mean is if you ask a suggestion, then he will suggest one of those sizes for the whole range. It's independant for every street tough.

But you have a couple of options too. For example, you can give him a hand too and he will tell you the EV of the hand for different sizings. But if you ask for a suggestion he will suggest the size he plays with his range, even if the EV is lower for this particular hand.

I don't know exactly how it works. The thing is it's not like PIO where you give options the program. It's very simplistic, a bit like a Nintendo game. You see the players, and you pick the action for them. But you can see their range and response for any action you picked at any time.

But when I say it wanted to bet 25%, it's because at any point you see also their recommendation, and it uses only one sizing all the time. So on the flop MP recommends to bet only 25% pot. But I could just make it bet 75% and then see his range and BTN response.

But then I know if I make it bet 75%, it assumes it could only bet 75%, because I can see the whole range is there either in the 75% bet range or in the check range.

Actually, when I say it picked 25%, this is the favorite one. But I can see that you can consult the range for half pot (bet 35.4%) or pot (bet 13.7%), or 2x pot (bet 1.3%). (for 25% bet 41.3%).

I think this is because these are the options snowie give itself to make a sim, but ends up selecting one size. So you still can see how he would play his range a different sizing, and I guess he select one of those options for the whole range.

Actually, it reminds me I saw an interview with Sauce (I think it was with Brandon Adams on youtube). He said something like a 'weakness' of many solvers, is they solve for some bet sizes, but if you bet against them something in the middle of 2 sizes he solved for example, he will just average something, instead of finding the actual solution.

I might say it wrong, but I think it's good enough to understand. I think that is what happens when I asked snowie to bet 75%.

But yes I will solve it in PIO using snowie's range !

I just entered a sim again, MP open 2.5x, BTN call, fold fold : Kd9s7s, cbet 3/4, Raise 1/2 (2.66x), and MP fold 64.7%. But if MP folds more than 53%, BTN can raise any 2 cards (unless blockers change the %). But BTN folds JJ TT AQo AoQs, AsJo AJo ATo... and 88o, 88s 78%... Overall BTN folds 50%.

I am sure if BTN 'knew ' MP strategy, he would raise all of those weak hands instead of folding them. Then of course MP would adapt, making it incorrect but still, the ultimate equilibrium would have MP fold no more than 53%.

To be fair tough, I 'forced' MP to bet 3/4 pot. Snowie wanted to bet 1/4. But he still can have a half pot bet and a pot bet range so it's not like snowie wanted to check 100% and I forced a bet there.

To simplify the question, there was a spot I think the way snowie plays it seems weird. But yes I give credibility to snowie. Otherwise I would not have a question.

But yes it seems to me, and again this is part of the question ( I love making things clear for everybody), it seems to me that mimicking human neurons is likely to miss the point of GTO, and yes it feels like it's called GTO to suit marketing purposes. But it is still an extremely interesting program, for which I remain curious :).

Hey also it's totally fine if you point out something I don't know * I just want to make it clear :). And if you tell me something I already know it's okay too :) I won't get mad :)

I don't understand what you are trying to say. I seem to not know something.. if I know something... if the solver is ''real'' ? What ? I was the one asking a question. Of course you're free to assume I don't know. But do you have something to say ?

If you think there's something I don't know about GTO or snowie you can explain it. That's the point. You seem like you are missing the point.

I can't look into a specific board now, but just look at the EV of calls when you run your sim, and compare it with sizings. For example the toy game where each player has either A,K,Q, when you have a Q if you bet more than pot you're losing because you just get called by A 50% of the time. So he means sometimes small size is better.

Thanks. That's true. For point one, when I was thinking about this issue, I was actually thinking about multiway pots and check-raising. I work with snowie for inspiration but snowie quickly is unbalanced as the hand develops. But I was considering a bit check-raise on QJ3 EP vs BTN/BB since I think BTN will bet AQ KQ QJ QT (maybe JJ I don't remember) and bluff some 88-99 ATs. I think if I check-raise at all the goal is to make Qx indifferent since he check back TT & Jx. But it's true that I also must be careful to not isolate myself against QJ and consider the next bet.

I think a common mistake that I catched myself doing with this toy game is to arbitrarily put a starting point. For example, it seems to ''make sense'' that OOP would bet A-K-Q and 5 for bluff, or IP would bet AKQ and one bluff after checked to... but this is arbitrary and obviously wrong. I think we have to start with things we know for a fact and reverse engineer everything.

-I know that if OOP bets, IP must defend 2/3 (at MDF), because OOP weaker combo should never win at showdown so her bluffs should be 0EV bluffs.

-If IP call 2/3 of the time, then his bluff catchers should be 0EV. The weaker bluffcatchers being for example 9-T, betting a J should (or could) then be a ''value bet'' (even if losing to A-K-Q).

-If J is a +EV call or a 0EV call IP this I don't know yet.

• I know that when a 6 check back IP , he has some EV (well if OOP doesn't bluff all her 5) . So, when OOP defending from a bluff, he has to overfold to MDF to make a 6 indifferent. If a 6 is indifferent, it probably bluffs sometimes. A 5 would then be a +EV bluff and bluff all the time.

-But how do I know that OOP doesn't bluff all his 5, and what is her value range I don't know yet. And I also don't know how thin IP can value bet but it can not be only A-Q, because then it would be ignoring OOP having the opportunity to bet before.

-I also know that after OOP check, she definetely has a range full of T-9-8-7, at least some 6 (because a 6 can bluff as well as a 5 because it will never check down an win). I guess the T-9-8-7 are ''bluffcatchers'' equivalent to the K in the K vs A-Q game, because they are not strong enough to value bet, but not weak enough to bluff. I guess then that when IP will bet after OOP check, his goal will be to make T-9-8-7 in OOP range indifferent (0EV?). So probably OOP must build a range which allow to defend a little less than 2/3 of the time (how much exactly I don't know but probably could), but then does she really need to keep strong hands into her checking range or could she just not randomize her calls with T-9-8-7 ? I guess if he never check a strong hands, then IP can value bet as thin as a T.

OK, I'm taking a look at the thread. However in my game it seems that J would be a clear bet IP because OOP's X range is weaker than initial range.

Ok but if both players had one of 10 cards from A-K-Q-J-T-9-8-7-6-5 I think it would be break-even to bet or check a J for 1.22$ EV, because he would get called 2/3 of the time, by A-K-Q, and 2 combos under J so maybe T-9. Then OOP EV would be , (3/9) -1$ + (2/9) + 3$ + 4/9 + 2$ = 1.22$. But when OOP would check, he would face a bet from ... euh.. I don't know .. still working on it.. I don't know because I would need to know the OOP XC frequency but the book doesn't say. Well it says the answer. But I don't know how to calculate it.

I understand that checking down a 6 allows IP to win sometimes. SO OOP must XC less than MDF.

so it would be 0.66$ vs 0.50$

Oh sorry the pot is 2$. so no betting a K would have an EV of 2$(0.5) - 1$(0.5x) = 0.50$

Hi, 1/3 of the time that IP check, K win. because IP will bet all his A (50% of his range), and 1/3 of his Q (1/6 of his range), so the other 2/6 of the range is Q. But yes I understand, betting a K would be 0EV. 50% win and 50% lose.

Hi, I would be happy to join. I play 200NL-400NL on a small fishy site and study PIO solver & snowie.

There is something that bother me. Again with the toy game with both players having A,K,Q, pot being 2, and 1 bet of 1$ allowed from either players, no raise.

I now OOP check all his K, then he has an EV of 0.66$, because IP bets all his A, and 1/3 of his Q. SO his EV comes from the pot being check-check 1/3 of the time (the 2/3 of the Q doesn't bet). This is pretty instinctive and I'm sure is the best option.

But I just happened to think, if OOP bets his K. Then 50% he will win 2$ (when IP folds his Q), and 50% he will lose 1$ (when IP has an A). So his EV would be (0.5) 2$ + (0.5) -1$ = 0.5$. He still makes money by betting his bluffcatcher. While it's a mistake, it's still a better mistake than if IP would bluff 100% and you keep calling at equilibrium, then your EV would be 0$. You would call 1/3 of the time, assuming you checked your A all the time, and when you call your EV would be (0.5) 3$ + (0.5) -1$ = 1$, but only 1/3 of the time. Then the other 2/3 you fold and have 0EV so total 0.33$.

If however your range would not have an A (still OOP) , and you would call 2/3 of the time with your K, well then your EV would just be 0.66$ vs an opponent who bets all his A and Q (unbalanced), which is actually the same as if the opponent would play balanced.

I find all of these things pretty confusing. At equilibrium it's not so bad. But I find it confusing that sometimes a mistake don't give EV, and some mistakes take more EV than others.

It's harder to punish capped ranges when they have position on you.

Hi, is it really harder to punish capped range when out of position ? If you take the toy game where one player has A or Q and the other player only has K, with one round of betting, I think the EV would be the same OOP or IP. Only if IP is allowed to have A, K, Q, vs A, K, Q, the EV improve IP ?

In real game since we are rarely totally capped I understand IP usually have an advantage everywhere, but I'm not sure if really OOP benefit from having a polar range (compare with being IP).

Also I would usually think the position makes the most difference on boards where the next card will change the board the most. That's why I think usually in PLO the position is probably the game where position makes the most difference.

But in reality, it is kind of the same thing happening both ways. You either should have KK-TT vs KK-AA and bluffs, or you should have QQ-TT vs QQ-AA and bluffs. In both ways I don't really know but I think it will be about calling at the right frequency from a GTO perspective and it sucks. But you probably have more EV with KK-TT vs AA-KK than with QQ-TT vs AA-QQ because a bigger proportion of the range is mergy.

I know in doesn't sound so sophisticated to make a strategy from 'brick flops' but with this range, all 9xx and lower are brick flops.

I understand and you may be right. I just mean if you look for equilibrium, if your opponent range would be like TT+ and AK only, and your range would be TT-KK, AQs and AKo, then he should not really try to value bet TT-JJ on bricks or only small bet small. If he does, then flatting KK becomes better.

Maybe his best strategy on brick flops would be like bet small a lot, then on the turn polarize AA, KK and AK a lot with big bets and check the rest (with a few slowplay). Or bet polar on the flop a lot with AA, KK and AK, and check the rest of his range with a few AA checks.

But I think if for example you would not have KK in your range, then his strategy should be to bet big mostly QQ+ and AK, and maybe there is not enough bluffs I don't know what would be the right ratio value and bluff. Or maybe he needs to use a few TT for bluff.

And maybe even AA. It's good to block Ax too if your opponent is the type to check Ax.