I know that the idea of "mixing it up" has been refuted, but I'm wondering if it's possible to do this in some kind of practically un-exploitable* way.

What I mean is, say you have this kind of strat when you're in the BB, BTN opens and SB folds:

- Start with the assumption that you flat everything (except for the bottom 5% of hands or whatever), but then ...
- 3b 66-AA, except for when you have, say, a and a
- take all other "groups" of hands and assign appropriate 3b frequencies to them, "randomizing" by picking a particular suit. So like, since you don't want to be constantly oop with the initiative with low-card hands like 64, 75, etc, you only 3b those when they're and . You'll want to have a higher amount of sd-able unpaired high card hands like KJ and AT in your 3b range, so you 3b them with more combinations, like and

Something like this. I haven't thought about this on a particularly high level, but the general idea is interesting to me. Obviously you can take this idea and apply it to postflop play, as well, but that's certainly more complicated.

Technically this is exploitable if someone knows you're doing this and they get to see you showdown in spots, but ain't nobody gonna have time to be figuring that **** out. And obviously you can change the suits whenever. I don't think people are every going to be able to figure this out and apply it in practice.

*This is what I mean by "practically un-exploitable".

To me this seems like some kind of weird combination of exploitative play and GTO play. Amiright?

You have a mix strategy but you haven't done any math or reason why it practically good for it to be GTO. I mean I think only bot can have a really good mix strategy since they truly can randomize unlike human where your emotion going to affect your state of play. I think you are thinking un-exploitable as in since people can't quickly figure out your mix strategy it must be GTO but in reality your strategy already sub-optimal.

You have a great poker mind no doubt and you obvious take different approach but you first need an understand of game theroy. I really recommend reading both of Phillip Newall since he make it super easy to understand. The second book is harder to digest but the extra information in there separate the decent regs from really good player. Some of the concept in there kind of what you are already thinking about but instead of pre flop it talk about board texture.

Also, the newer GTO bots are almost completely eliminating a mixed strategy. They are only using a mixed strategy with hands that are right on he borderline between call and raise.

But ya, a perfectly balanced 3-betting strategy out of the BB should be better than always calling against a lot of opponents... especially ones who check back flops really well. In fact, always calling really sucks against someone who checks back a lot of flops hu.

As far as preflop play is concerned, at least, I don't know why emotions would play a factor if one were to stick to this formula.

You're definitely right in that I haven't done any math on the subject, but I think my ideas are at least sound theoretically. Though I fully admit this whole thing might be dumb for some reason I haven't reached yet.

I've flatted 100% for many years now, but have always suspected that having a balanced 3b range is best. That's the impetus for this idea.

With that said, I'm not getting any non-terrible opponents that are checking back a bunch of flops on Bovada. Is this the new hotness with good regs?

Agree totally with first part. Mixing should the way i see it only occur super rarely when you are at the exact threshold. So to just never have a mixing strategy with one particular hand is probably a good idea.

But somewhat disagree with second part.
The strategy i think is best (for a human with limited computing powers anyway) is to always call from bb but then have a decently sized and balanced donking range vs those who check a lot of flops. (so most regs these days, myself included)
About always flatting i think its marginally suboptimal at worse in isolation, but then has the added benefit of making making balance postflop a hell of a lot easier on a lot of boards.
Being OOP preflop is main reason why i think its worth to give up a little bit EV preflop, with the thinking that we get a very good chunk of it back from the flop onwards. We can simply pick our spots a little bit better by not pumping the pot so much so early.

Hen, I mostly use an always call strategy in the BB. But I think it really sucks against people who check back 20+% for example, which is getting more and more common vs. a button raise for example. And ya, I agree that figuring out wtf a GTO 3-betting strategy might look like out of the BB is extremely tough. We can try to extrapolate it from the GTO bots' hu 3-betting ranges, but still.

But as you know, it becomes more important to 3bet vs a wide range like the button vs. someone who checks back a lot as opposed to an UTG raise where very few people currently check back flops.

I think the difference between 3-betting and always flatting is a lot bigger than you are making it out to be. All good hu players know that an always call strategy sucks green dog dick huhu. If it is so horrible huhu, it has to be at least somewhat bad against a player opening 50+% on the button.

On donking... I have been doing this extensively for 50k+ hands and it is REALLY hard to balance and I am not convinced it is very awesome as part of an always flatting strategy. It is much more complicated to implement than a 3-betting pre-flop strategy imo. You now need a balanced donking range, a balanced c/r range, a balanced c/c range, and if you really want to go all out, a range where you wait til the turn or river too. If all of our aggression on certain board textures is from donks, our better opponents will know we have jack and **** when we check and will pwn our face. This happened to me for sure. Also note that most people who check back flops still don't know how to do it well and how to follow up on the later streets. If their check backs are very predictable (like mine prolly are when I am playing bad), then we don't really need to donk because we can just annihilate them on the big streets. And so on.

All that being said... I think a 3-betting strategy is easier to implement than a donking strategy. Especially when you consider that we don't need anything rsembling a GTO 3-betting strategy. If we know what board textures our opponents like to check back, we can easily adjust our 3-betting range to the hands that hit these board textures. And on and on and on.

LHE confirmed complicated and not dead... winrates down, games tougher. But still lots of room for regs to improve imo.

some times I just want to punish my opponent for raising my BB for the umpteenth time. So I 3bet!

Pretty sure it's perfectly balanced.

I actually agree with Hem regarding donking vs a btn that check back a lot. Regs are def. checking back now and days but they are doing it on the wrong type of board. donking range should be 100% about board texture. for example low/wet board should be check back more but sometime I see reg check back K7xr type which should probably be close to never in a 6 max btn vs bb since the 50% stealing range is still pretty strong. I also agree with Unguarded regarding not having a 3b range huhu is pretty bad but the situation is a lot def. 85% steal vs 50% make 6 max btn range so much stronger our 3b in BB will give up too much information. In some way I think of it like counting card in BJ low card are good for (23456) your range middle card are good for both(789) high card(TJQK) are bad. The A card is a little bit tricky and it depend on the texture. I will use this to determined if after I c/r flop or donk flop if I want to cb the turn or check/call and check to give up sometime.

Its pretty much impossible to prove mathematically wether coldcalling 100% or not from bb is best vs steal. But i remember a couple of years ago the empirical evidence i saw from several winning players with different styles over millions of hands showed that it was basically a tie between the two different strategies.
I chose to coldcall 100% though because of a couple of factors:
- I think it is then easier to get postflop right because you enter the flop with a wider range. Both in terms of balance and exploitation.
- Since donk checking flop after preflop 3bet is likely a dominated strategy it means you really are setting yourself up in a vulnerable position being OOP. I think this has worked so well in the past because players generally have been really bad playing without initiative. Like, they either fold too much or they feel obliged to raise too much with weak value, because someone told them initiative and agression is good by default. This has changed a lot last years though so i feel if anything coldcalling is likely better now than it ever has been.

HUHU is ofc different because ranges are wider so if its close in BB vs BTN it should clearly be correct to 3bet HUHU.


Also i definitely disagree that its so hard to balance a donking range. No reason why it should be harder than balancing a c/r range.
Like DJ said it is 100% board dependant. On some flops vs some opponents i would never c/r and always donk my "agression range". On some flops it would be the opposite and i would never donk.

If I had to put money on higher winrate between some unbeatable bot that only tried to be unexploitable vs. the best exploitive human players, I'd bet on the bot, at least at stakes that matter.

It is a property of GTO strategy that every individual play within the strategy is optimal. For example the GTO BB defense play with T8s will also be the highest EV play with T8s. A strategy where we sometimes "sacrifice" EV from T8s to make (say) AA-TT more profitable cannot be GTO.

One way of thinking about this is that the GTO play is still the GTO play even if this is the very last hand ever. GTO strategies cannot rely upon future deals to make up lost EV.

Mixing cannot be GTO unless two plays are exactly tied. Even if they are tied it doesn't matter which you choose and mixing is no better than not mixing.

It is also not necessary for Villain to "guess" your non-GTO mixing strategy in order to exploit you. All Villain needs to do is play his GTO strategy and your non-GTO strategy cannot gain. Your EV will be less than or equal to what it would be if used GTO strategy.

Of course if you play a non-GTO strategy it is likely that Villain has an exploitative non-GTO strategy that will hurt you even worse than his GTO strategy does. I'm not saying that GTO strategy is Villain's best option for exploiting you, merely that it is good enough.

The technical term "exploitable" may be unfortunate for teaching purposes. It conjures up an image of Villain coming up with a diabolical counter strategy that crushes our flawed strategy.

That can certainly happen but it is not the only way.

As a practical matter Villain usually "exploits" our bad play by just showing up. He does whatever reasonable thing he normally does and we fall flat on our face. Our flawed strategy throws money into bad situations or leaves easy profits sitting on the table.

There is no conflict between this statement and the idea that always calling may be GTO.

The theoretical premise behind always calling is that opener *should* always bet the flop.

Why should opener always bet the flop?

Opener's range is much stronger than defender's range. Most of opener's hands are strong enough that they should be bet for value against the defender's weak range.

The remaining weaker hands should be bet as a bluff. You can view this as balancing your value bets or you can view it as your strong range makes it +EV to bluff. Two different ways of thinking about the same thing.

The premise that opener has the stronger range is important. This is a very strong assumption if Villain is a normal player opening UTG. It's a reasonable assumption if Villain is a LAG-TAG button. It probably breaks down in aggressive HU games where either player can have any two. It is definitely wrong if Villain opens any two but limps with all good hands.

The hand range strength differential is what makes always call work. Don't leave home without it.

So what happens if Villain ignores this post and checks back a lot of flops anyway? This is an example of my last post about how bad plays get exploited. Villain falls on his own face. He loses so much EV by not making continuation bets against your normal hands that you can more than afford to give an occasional free card when you have a big hand.

Is always call the very best way to exploit a frequent check-behinder? Probably not. But it is plenty good enough to hurt him really badly. What we do against this player has nothing to do with correct play. It is merely the timeless question of how best to attack a Villain with an obvious leak.

Corollary: if Villain is checking behind because you are an always-caller, then he is only hurting himself and your strategy has gained an undeserved extra benefit.

Are you saying that c-bet 100% otf when hu is GTO?

There's got to be situations where checking behind the flop is the most EV. And it seems hard to believe that betting 100% of the time is GTO in any situation. Especially vs a non 3 bettor, there's flops where we'd either be behind or even in equity, will get c/r a ton, gain little by betting but gain from a free card.

I have no real evidence to back this up by my btn WR is very good and it just feels right sometimes intuitively. I find I'm able to regain balance by the river (rarely raising the turn after checking behind flop)

I agree with all of this. Also, I think that having a good check back strategy allows for a wider stealing range. More hands for a profit = higher winrate.

Actually i am 100% certain that GTO play involves having a checking range on flop as BTN stealer.
The only spot i can recall where betting 100% of flops is likely good is when we are OOP, as in BB 3bet vs late steal HU f.ex. Multiway i guess GTO would include some donk checking even in that spot though.

GTO bots most certainly make individual plays that sacrifice EV to benefit their overall strategy. A simple example is that every GTO bot I have ever seen will 3bet 54s huhu and a lot of other odd hands like J5o, Q5o, 65o, etc. depending on the bot. Additionally, they will just call with some hands that are clearly worthy of a 3bet like A7o, A3s, KTo etc.... again to benefit their overall strategy even if they sacrifice immediate EV. And on and on and on.

If it is the last hand, 3betting 54s, 65o, Q5o, etc. and all the other weird hands GTO bots 3bet huhu would be very stupid. Just calling would be better.

This contradicts everything that has ever been written about GTO play and what we have witnessed in action from the bots. I am not goignt o go through MOP right now, but there are tons of examples of very simplified games where a mixed strategy is necessary to keep our opponent indifferent between calling and folding. Like say we are playing huhu LHE and the only hands we can have are AA, KK, and QQ. If it is not super obvious to you that we would need to be using mixed strategies in this game, then I just don't know how to help you. And the plays would not be "tied"... we would have to mix our strategy to reach the appropriate ratios. Failing to mix would leave us wide open to being exploited.

At least this part is correct.

It is not merely "good enough" in actual play where there is rake. Say there is .5 BB/100 in rake and a world class reg is playing a merely good reg. It is very possible that if the world class reg plays a GTO style, he will not be able to overcome the rake. But if he plays an exploitive style, he will be able to overcome the rake. But yes, a perfect GTO bot would obviously crush online in practice. But a perfect exploitive bot would crush much harder if it had a reasonable sample size... or if it were capable of playing perfect GTO until it found leaks in its opponent's game.

Checking back flops completely owns an always call strategy in huhu which is why every world class huhu player and bot 3bets huhu. Yes, tighter ranges make a huge difference (such as UTG in full ring vs a BB call).

Anyway, I get the impression that you have little to no experience playing with or observing top GTO bots or playing in tough high stakes games on Stars. The little things really matter nowadays when you aren't in super fishy games. And even in super fishy games, the little things still make a huge difference in the long run since we are sometimes playing pots with just good regs. Many of us play in an environment where we can no longer consider good, but not ideal strategies to be acceptable. We will never play perfect, but we have to try and cannot settle for "good enough".

I would assume GTO strat HUHU dictates that 3b pf is correct, under the assumption SB is chking back flops at a "correct" ratio. ( You know this anyway)

There are still villains at decent stakes HU 30/60+ that dont chk back much for various reasons (many of them correct)

There are many variables to optimal pf play but you are obv correct that SB playing "GTO" will require BB to respond with a 3b pf strat

There are myriad reasons why cbetting 100% at 6m may be more correct than a human representation of a chk back strat. The range vs range thing may or may not be relevant depending on factors. The strength of our range pf is often a reason we should not be cbetting. So yeah its not as simple as some suggest (eg our pf range is way tight= always cbet)

As usual the answer lies somewhere in the middle, but from my observations ($30/$60 and below) given that its somewhat a zero sum game, spending time perfecting an exploitative approach to the game will probably be better time spent than perfecting a GTO style approach. There are still plenty of "mistakes" being made by good regs, although what they are worth and whether they can be exploited is hard to say as i havent been playing much last few months.

So im not wanting to sound like an authority on it.

However, if you feel that 6m games will end up closer to an equilibrium strat, we probably should spend more time on a more balanced approach. However, i doubt it will ever get anywhere near close + Humans suck badly at applying GTO at HUHU, let alone 6m

ok what is GTO means again?

It might be "better" depending on some factors but it's not GTO. The point of GTO is to be unbeatable, even when facing a sick-soul-reader. If you would alter your 3betting range on the last hand a sick-soul-reader could exploit that fact and take advantage of your strategy.

Of course a perfect exploitative bot would crush harder than a GTO bot. The thing is though that we have a mathematical proof that a perfect GTO bot exists and a decent idea how to create it while we have almost no clue how to construct a perfect exploitative bot, even in theory.

troll or not, here I go:

Game Theory Optimal. Meaning, a set strategy that can not lose to any strategy, including itself.

In practice it means that for every situation you know exactly what to do, regardless of your opponent. His tendencies have no effect on your strategy because it can not be beaten by any set of tendencies. In fact, you don't even care if your opponent knows exactly what your strategy is. Your plays are mathematically guaranteed to be at worst 0 EV and can not be exploited.

As long as your opponent doesn't counter you with a GTO strategy, he will make some mistakes and lose EV. If he plays GTO you guys break even (minus rake).

The general consensus is that humans play so far from GTO that a GTO strategy would crush everyone for a huge winrate.

there's no such thing as "under the assumption that the opponent is doing x" in gto strategy. gto strategy will not change no matter what our opponent does. as soon as we adjust to our opponent's strategy we are deviating from gto towards exploitation

no. the one and only thing it will require is for bb to play gto as well (whether or not a gto strategy dictates 3b pf I have no idea, so in practice it may dictate a 3b pf strat, or it may not), so the sum of their game will always be 0ev, any other option will be either 0 or -ev, which is worse than guaranteed 0 ldo.

what did I miss? where can I see that proof that such a bot has been created? or by "exists" you mean it can be created theoretically?

Sorry, you are right. Im using terms way too loosely + I have a very basic understanding

What i meant to say in relation to UG's post is a GTO strat will involve 3betting some % pf . Yes this never changes no matter what villain does, its a static strat. If a human player is striving to play a GTO type strat, he may win less by continuing with the same pf 3b% vs opponent x, rather than shifting his pf ranges slightly

Same re: BB vs SB statement i made.

If human BB is playing vs someone who is playing GTO in HUHU, he will be best served to play GTO also, which will include a 3b strat pf. Didnt mean he should just incoporate a 3b pf strat in isolation

The point im trying to make is we are wasting so much time on GTO strats in 6max when the game is so far from what i imagine 6 GTO bots would play. I mean you still see guys trying to play GTO vs semi fish and terrible regs that have glaring postflop leaks, its laughable