Tombos, what are your final thoughts on this?

Especially back to my original question of what would the benefit be for solving with 4+ sizes on the flop when the overall frequencies and EV don't seem to change compared to a simplified strategy?

If the EV is exact same that means that assuming you can execute both these correctly (and of course your opponent too) choosing either strategy won't lose any EV. So in other words in theory these 2 strategies are equal.

I'll let someone else make the argument for more sizes but i'd be perfectly happy to work with your simplified sim. The main argument of course being: execution

Great question @newguyhere.

Firstly, I don't think you need 4+ sizes in most spots. You just need to faithfully represent the gamespace such that the solver doesn't significantly exploit the limitations of your tree. That often (but not always) means needing to include multiple raise sizes, donks, 4x overbets shoves, etc.

This is just my subjective experience, but oversimplifying the betting tree often tends to overcomplicate the strategy later down the road. Conversely, extremely complex trees can lead to more natural strategies on later streets. I believe there's a simplification tradeoff. But again, that's a subjective opinion.

Now your specific example seems to show that there's no real strategic difference in that spot. But I'm sure I can find counter-examples where the strategy would change drastically on the river (comparing very simple and very complex sims). I might post some examples to the theory subforum later, but for now I'm exhausted lol.

I think there is absolutely a "sweet spot" in how many sizes you need, but that sweet spot is not well researched. The underlying theory of abstracting gamespace, in general, is not well understood.

I think it wasn't included because it assumes your opponent is playing gto and not adjusting. It would be a fine measurement against a bot, but not a good simplification vs humans. It's much better to lock your strategy and resolve the entire tree, to see what the loss would be if your opponent is allowed to play any strategy.

It’s exactly **** like this that has made poker so incredibly boring, and why there’s basically no fish left. All because people are so desperate to prove how smart they are. Yawn.

You quant nerds have really made huge life blunders dedicating all this mental energy to a card game when most long time winners know the name of the game is max explo vs the fish and most regs. Meanwhile you nerds dive deeper in the theory to look for some supposed edge you will gain, delusion level rivaling many in academia. I know in your triggered state you'll cite me some HS crusher who you look up to; way to leave out all the people that tried an failed approaching poker the way you all do. Your classroom quant approach falls apart for a vast majority of aspiring nerds trying to make it in poker once they get punched in the mouth a few times via the incredible unpredictable nature of poker.

I look forward to seeing this view more widely agreed with, IMO it already would be if there were public lists of all those who tried an failed, but sure the makers of these apps can keep pushing the scam dream that hey if only you study these sim solver inputs just a LITTLE more you'll start finally beating the games!

Having this discussion in NVG is the equavalent to 2 'pros' discussing high level strategy at a live table. The main result is recreational players feeling like they have no chance to win and being less motivated to return.

I think for the vast, vast majority of serious players at midstakes or especially lower the value of spending time figuring out complex stratgies like being discussed in this thread is at best much lower than your hourly playing but more likely an overall net negative as your execution is going to be awful.

My gut feeling is people selling this stuff (with the exception of high stakes) are snake oil salesmen drawing in non-winning players with promises of the 'secret sauce' to winning.

Or maybe, and this is going to sound crazy, some people enjoy learning about theory.

What we're seeing today is no different than what we saw 10-15 years ago when all the old school players thought the young internet kids would go broke.

who are the young internet kids going broke and who are the old school players in this argument?

Grunching. Is this really about people beefing over simulation algorithms?

That's my point.

The old school players thought the internet players would go broke, but they didn't.

This is the 2021 equivalent of "HUDs are killing poker". Truth is most regs play nowhere near GTO and the average recreational doesn't know or care about it.

Government regulations/segregations is the main reason why there's basically no fish left. And I'd argue that there's still plenty of fishes left.

Who cares. What this type of approach has done has turned poker - a fun gambling game with an edge - into a boring, robotic, computer solving maths lesson. **** this ****. No rec, rich whale or anyone with any real social skills or success in life wants this type of discussion. They might not read this or even know about it - but it’s the approach to the game by geeks has destroyed the game. This is just my opinion. I’m not alone, every wealthy rec or whale I know has either given up, or plays in private games where no socially inept nerds are solving for every outcome. Zzzz

exactly.

This thread was actually very interesting. First strategy thread on a forum that i followed in ages. Good stuff.

Some people apparently like it that way and that’s their prerogative. You play/study the way you want, they play/study the way they want.

I enjoyed following the discussion.

As for my own thoughts on this topic:

I have been going back and forth on complicated vs simplified strategy/game tree. Now after reading this thread I'm thinking it is good to study both methods, both simplifying the complicated game tree and using a simplified game tree, but in practice at the tables you should more often use the simplified game tree and try to execute as optimal as possible, especially vs opponents you play alot with, because otherwise over a certain amount of time likelihood increases of your opponents adjusting to your flawed frequencies and exploiting you, if you use the simplified version of the compliicated game tree.
However if you study more complicated game trees, from time to time maximizing ev of certain hands has to be a good idea too, you can do this even vs players you play alot with if you only do it occasionally. That way it is impossible for them to adjust.
Of course all that is assuming that in any given spot we have good ideas about various solved game trees.
This reminds me also of this video by Daniel Dvoress on Runitonce of a flop spot which goes into the topic of simplified vs more complicated strategy as well, where Dvoress makes the case for a more complicated strategy https://www.runitonce.com/poker-trai...ng-large-bets/
I made the following comment on that video which i think fits for this discussion as well

"regarding loss of ev by following the simplified/worse strategy with the 100% small cbet vs the more complicated big bet/small bet/check strategy: Is there a way to figure out how much ev you can lose by following the more complicated strategy, but not mixing correctly (using big bets too much for example, or checking too much) and where the threashold is, like how accurate do you need to be with that in order to really gain ev over the small cbet 100% strategy, since that one is easy to follow and you cant mess up at all. I think that factor makes it not so easy to say that the more complicated strategy is better for everyone. Also i would actually bet that nobody can mix the frequencies perfectly there as used by pio. Thats why i think its important to know the margin of error that allows up to still gain advantage over the simple strategy"

my favorite thing about dumb posts and bitter posters like you is that you can't accept that people who are interested in the math behind the game also have social skills and can have a good time at the tables. Laughably ignorant. More likely you are the insufferable one.

"every wealthy rec I know has moved on"

LOL

It was indeed a good thread and i really enjoyed the discussion over a concept that i‘d never seriously looked at because it seemed so foreign to me. That i basically dismissed it right away.
I thought you guys we‘d be tired of it at this point but I’m glad you enjoyed the discussion so I’d like to continue it. I‘ve sent tombos a pm because i didn‘t want to clutter the thread originally but since we are all enjoying the discussion i might as well make it public so we can all learn from this.

While I am in full agreement with Tombos over the toy game we created I question whether that toy game is actually applicable?

By assuming villains GTO strategy stays consistent we are ignoring the effects card removal has on our villains range. Put simply. If nothing else changes about our range, yet we suddenly have 4x as many top pairs our opponent will have top pair less often.
How do we account for this? If we just pretend this ain't a big deal I'm happy to dismiss the entire concept right away. Atleast for bucketing large group of hands like top pair.

As for the other debate going on right now about sizings:
it's my opinion that almost everyone way overcomplicates the game tree and overestimates what they can execute. Including myself. People like to do research into sizings and while I can't technically fault them for it they are chasing the tiniest bits of EV. With 0 hope of ever executing their strategy in real life. I can assure everyone there is a lot of low hanging fruit for you to pick up if you study with simple game trees.
It's also my opinion that you can't run a 4 sizing solve on the flop and then simplify to 2 sizings without running a resolve.
If you only want to use 2 sizings that's okay but you better solve for those 2 sizings. Same thing if you just want use one. The resolve needs to happen though!

I think of running sims this way. If my biggest leak is only betting 75% pot in a spot where I should be betting 100% pot I'll be fine. There's no real EV in knowing which size is correct. There is however EV in knowing how to play your own betsizing strategy.
This is true(IMO some people will disagree with this) as long as you avoid straight up blunders like not allowing either player to get all in in certain nodes for example.


edit:
This seems to be a never ending debate but I'm fairly confident in my take. i'm trying to find a post by Scylla (who's the creator of gto+) who pretty much agrees with what i wrote regarding amount of sizes.

I've seen you say this a few times, and I'm wondering has anyone disagreed with you?

Seems like common sense for anyone with experience using a solver that if you are going to change any sizes that are used, or how some hands are played within each range, that you would have to resolve.

Its inevitable though there's no point blaming the people who are using every edge to improve. If you replay a Chess game from 100 years ago it will be littered with theoretical mistakes by todays standards. Have never used a solver in my life and agree its been terrible for the game but you can't stop progress.

Oh no. No one on here has disagreed. It really is that obvious. Once you change anything in the solve whether that's sizings or strategy for certain hands you need a resolve (!).
Finding equilibrium disagrees however. His entire product is based on 4 size flop solves.

He even admits he only uses 2 sizings most of the time but doesn't bother rerunning the tree. FWIW this is literally the main reason I called GTOX worthless. It's that flaw that breaks the entire program for me.
Basically GTOX follow a strategy which assumes a 4 sizing split. A strategy not even the creator follows. A strategy I don't follow.
So exactly how is this programm useful for ME to analyze my hands?

I'm not prepared to die on the EV regret hill although I strongly question how accurate of a measurement that is still due to the valid concerns of frequency (which is undeniably true) and card removal (which i am not smart enough to comfortably say whether it's true or not). That I also feel comfortable dismissing that concept.

by the way I found the scylla post I was referencing earlier and since he is the actual creator of a solver I'd expect his words get some respect:
https://forumserver.twoplustwo.com/s...ostcount=12088

There's another large advantage mentally - if you are actually confident in your strategy it is very helpful in dealing with negative variance

There is no additional blocker effects whether you hold optimal or 4x time of top pairs at least not for hands that mix in real gto.
KsQc always blocks same hands in opponent range regardless of how often you have that hand in particular spot.

Of course we block the same hands. That's obvious.
In other words you say whether you have KsQc 1% or 100% has no effect on our opponents range?

won't he be more likely to hold Kx himself if we ourselves have a King less often?
so to go with an extreme example since we are long past the point of playing GTO:
what if our range is literally only KsQc when we do bet? Do we still pretend they can have KsQc?

Edit: I'm fully aware we discussing some abstract theory here that's not relevant to the real game. It's fun though and the people enjoyed it.

Also on the off chance I am wrong about frequency mattering when it comes to blocking stuff i'd reallyy like to learn.

Certainly. This has been my mindset while grinding hr mtts over the past years.

Yeha it doesn't matter whether you have KcQs 1% or 100% it just determines how often you are in particular spot with particular hand.

Well if you hold KcQs you don't pretend they can have same hand even if you have bunch of other hands. I mean that the whole point about blockers.