GTO wizard recently released reporting stats for their "complex sims". I thought it would be interesting to graph the cbet stats. There's a strange trend that comes up in UTG-BTN vs BB single-raised pots:






It seems to prefer small/medium cbets, sharply dropping off after half-pot sizes, then a small resurgence around the pot+ sizes. Why does this pattern exist? What's with the disdain for the 3/4 pot kind of sizes?

most used sizigns are probably the ones that make most of villains range indifferent. I assume not many hands in BB defense range happen to be put in a tough spot with a 3/4 bet

Why not?

In a single raised pot the 3 street geometric sizing is ~120%, so it's no surprise to see the solver cluster around that size when it wants to polarize. The reason the polarization bump is a thing to begin with is because of the asymmetry between attacker and defender ranges, which makes it such certain flops play out very similarly to the AKQ game.

3/4 is simply not an efficient size at this SPR as it is not large enough to use in pseudo AKQ situations and too large for depolarized betting.

Interesting. Yeah the geometric sizing seems like the best explanation so far for the bump at the end. Though I'm quite surprised the 50-90% zone is so low.

I might need to update my cbet sizing strategy.

I got some more charts from BTN vs SB 3bp:





a couple observations:

1) SB has NO chill, 80% cbet OOP seems really aggro to me

2) Geometric bet sizing is 56% for this spot. We do see a sharp decline after the next highest bet size.

3) This curve is super satisfying

I decided to spam this thread with the rest of my BTN vs BB srp bet sizing trend analysis. Hope some of you guys fine it helpful!

Errr on the other hand, I probably shouldn't have shared all this paid data. I was so busy thinking about if I could, that I didn't stop to think if I should.

Mods, feel free to delete this thread.

I don't understand this. You made the graphs right?

Sharing data/information is a good thing.

I did yeah, but I didn't make the sims. I've never had access to bet sizing data of this resolution so I was excited to explore the trends. But I'm not sure about the ethics of sharing the resulting data?

IMO your ethics are up to you. My opinion on IP is that I’m heavily against the concept at all. I wouldn’t feel a shred of guilt pirating your grandma’s cookie recipes.

A couple of factors are at play here. First of SB's range is probably too strong since I assume this was solved with 3bet or fold preflop ranges. Second polarizing too much when OOP really sucks because there are still 2 streets of action left to play with positional disadvantage. The goal of OOP in these situations is usually to try and delay the polarization until the turn. We all know how much it sucks to be OOP and to check and face a bet. This is what the solver wants to avoid if at all possible. This is why you see when given the option OOP will use even bets as small as 10%. It's just trying to negate positional advantage as much as possible.

Don't worry, this is fine. Unless you would share the full reports, we don't have issues with that. Glad to see you are enjoying our Aggregated reports!

Phew, that makes me feel better. Thanks!


SB is allowed to have a calling range in these sims, but due to the rake structure, it's virtually 0, so to your point about the3bet being very linear stands.

It's certainly cbetting aggressively, and you might be right that this comes down to the linearity and the positional disadvantage. I wonder if we'd see the same result with a more polar OOP 3bet range, like BB? Or SB 3b vs later positions? I'd have to upgrade my subscription to access those reports, but I might just do that to satisfy my curiosity.

Here's the most interesting question as far as I'm concerned:

1) Why are the IP vs BB ranges bimodal?
2) Why is the SB 3bet vs BTN range so much more uniform, as compared to 1)?

After discussing with some really smart people, I've arrived at the following ideas:

There are two major factors involved:

a) The SPR - The geometric bet size is about ~115% for the single raised pots. It's closer to 56% for the 3bp. So we see a bump around that size for each.

b) The range asymmetry - The BB has a big discount to enter the pot, so when we observe a spot like BTN vs BB there's a massive range asymmetry. Conversely, when the BTN is facing an 11bb SB 3bet their discount to enter the pot is much smaller.

Warning - wild unscientific speculation ahead

The bimodality of the results likely indicates that the solver prefers either a smaller linear-ish cbet or a more aggro polar strategy on the flop. This bimodal effect becomes more apparent as the geometric bet size increases and is probably exaggerated by the range asymmetries. To what extent each of these effects plays a role i do not know, but I intend to find out.

If anyone has any other ideas please feel free to chime in!

Highly bimodal:


Less bimodal, more uniform:

Yeah, it would be interested to see aggregated reports at different stack sizings and even with different open/3b sizes. Maybe it would lead to deepened understanding or maybe it would lead to more wild unscientific speculation xD