Ty for the link the theory section is so far down I missed it. Seems like a lot of the confusion there stemmed from units, e.g., folding freq 3% above GTO ≠ defending 97% of MDF (58.2), the 1st scales with pot size and the 2nd doesn't. The later post about SD bias is interesting. If you had a decent hole cards revealed (so SD bias doesn't exist) DB, cloned it, and made it such that non SD hands are hidden you'd be able to measure the effect of SD bias perfectly.

My DB says they're betting 46% instead of 40%, not sure if you'd consider that 6% or 15% more than GTO. Their sizes are more condensed, running the numbers regs are putting in 28.46% of the pot on average vs 27.1% for GTO so it's kinda 1.4% or 5% too. The C-B line is interesting here because their frequency doesn't really go with flop size as GTO does, so C20-B is probably more than 10% for you. In GTOWiz after b33 or b20 your hand actually loses a bit of EV once your opponent bets turn for any sizing. They're overfolding pre and flop but it seems like they still manage to be too weak betting here.

Very cool hand, I'd never have jammed here even if I could see my opponent had QT

Interesting idea, to my knowledge no one has figured out how to quantify SD bias effectively. Software solving this puzzle would be invaluable but also make you a good amount of money.

Okay, results

His play seems very strange

I think it's pretty good, my line is overfolded.

Ok, I have to ask then my friend... if you understood that, then I'd think you'd understand that he'd have a lot of bluffs here. Again, going back to my first post, if you charted out the EV of calling vs bluffing w/ your whole stack, I don't think this one is very close.

Yeah no problem we can agree to disagree

Sure, of course we can. What are we disagreeing about?

I think your concept / theory is fine here, but for some reason you're minimizing the amount of bluffs your opponent has here to make this concept work. Let's simplify... what's your opponent's bluff : value / split ratio in this spot in your mind?

I thought you were advocating for a call? I'm in the jam camp.

MDA says it's 35weak and GTO is 30w so 65/35 would be a reasonable guess for me.

I'm not trying to minimize the amount of bluffs my opponent has, the more bluffs the better for me.

Yes, I'm in the call camp. Actually the lower that bluff / value ratio gets, the more you should be calling. You're trying to get your opponent to fold only QT and splits. The more bluffs he has in his range, the more the absolute EV calc between shove / call goes from neutral to negative because of how much you're risking. It's not just about how often he folds QT and splits to this line.

Aren't we missing a massive part to this equation? How often he folds?

When you say the lower the bluff/value ratio get's the more you should be calling. Why is that?

Of course how often he folds if part of the equation. It's one of the most important parts if we know it EXACTLY.

Look, if you're plotting this on a profit bell curve, if your opponent is almost never bluffing, then your only chance to win the hand is to push him off better hands or splits (because he's not calling w/ worse). But if they are bluffing at a higher ratio, then the ratio of your calls should naturally increase, as it will net you more EV across this bell curve between call / shove. If we knew for sure the exact % of our opponents fold%'s on this line, then it doesn't matter how much he bluffs... but we don't know that, unless you want to say your MDA data is perfect.

I'm glad you brought this up because this is something Tombo's just talked about in my other thread. Basically, you will never have perfect information in poker so what you do is assign a confidence level and then a standard error stat to the data.

So we were trying to figure out how reliable my X-B70F BBvsBU data was based on the sample size and he assigned confidence levels and then standard error stats based on how many samples we had. Here is an example:



Once I get the C-B-BF data we can do the same thing. It's a much more scientific approach and we can use it to see how likely it is that they fold over MDF.

I'm excited for the data.

All these numbers depend on flop and runout.

If you need 1k hands to have 95% confidence level for line. You need 1k hands for this specific texture too.

Agreed, we probably won't have enough of a sample to to conclude anything with a 95% confidence level, although we might with a 68% confidence level.

Not just dependent, critical. A paired flop w/ a broadway runout is going to be overbluffed a lot. A lot more than just some random line.

DDP, so what were the actual %'s you were landing on? 3% over for bluff and fold?

I don't have the fold data yet but it's overbluffed by 5% overall. Not taking into account board texture/runout.