Theory/Exploitative Play on K72r BTN vs BB
I was going deep analyzing this board and came up with a lot of questions around how to play it both in theory and in practice in 50nl/100nl. I don't have a ton of experience (<100k hands total), so hoping people who have more experience or who are familiar with MDA can help give input on some of my conclusions. I play on a site that doesn't work with PT4 or other similar software so it's very hard for me to get any data around this.
Firstly a few points on theory:
(1) BTN is supposed to cbet ~65% of range and mostly uses 1/3 sizing
(2) BB is supposed to x/r ~12%, and when they do, BTN is supposed to call ~60% of their cbetting range
(3) BB is supposed to only fold ~32% against the 1/3 cbet
(4) In x/b/c flop line, BTN is supposed to primarily use overbets, 125% or 175% on the turn when the card is a 9, T, J, Q, A
Some of my assumptions about low stakes tendencies:
(1) BTN is likely to over cbet this board
(2) BB is not finding 12% x/r, and is likely to be too value heavy due to not enough bluff combos (are people really x/r 76s or 98s or JTs here?) and also to slowplay too many of their 2p and sets relative to theory, and they're not x/r their thin value as often as theory
(3) BTN is overfolding to x/r, they're supposed to call all kinds of stuff like A8s pure, T8s bdfd pure, 86s pure, etc.
(4) BB is likely to overfold to the 1/3 cbet
Given the theory and assumptions, I would think some good exploits would be:
(1) As BTN, cbet range because overfold and lack of x/r from BB
(2) As BB, x/r way more bluff combos than theory and slowplay more value
(3) Overbetting the turn as BTN is not a good idea, since BB is folding a lot of the bottom of their range on the flop and slowplaying a lot, so we should use smaller sizing on the turn, maybe like 75% instead of 175% overbet
Any thoughts on this analysis or the assumptions I'm making here?
10 Replies
Depends on skill level of your opponents, but most regs in todays gams will XR 2p+ even good TP for value.
I would still XR value as BB even if they over fold you can size up a bit, same for cbet if they over fold and under KT+ loses ev so you can size up with those.
Turn if you think they wont XR value you can size down ofc
Nice analysis!
(1) As BTN, cbet range because overfold and lack of x/r from BB
Agreed.
(2) As BB, x/r way more bluff combos than theory and slowplay more value
Definitely bluff more.
Whether or not to slowplay is not so clear. Ultimately you want to know if more money goes in through the XR line or the XC line.
You can't be certain that XR < XC just because they overfold to a raise. Perhaps they overdefend vs turn/river barrel. Or perhaps they're too passive after you slow play.
Also don't forget there are extra incentives to raise vulnerable value hands for equity denial.
(3) Overbetting the turn as BTN is not a good idea, since BB is folding a lot of the bottom of their range on the flop and slowplaying a lot, so we should use smaller sizing on the turn, maybe like 75% instead of 175% overbet
Disagree - kind of. The decision to overbet is more about polarity than range vs range equity. Just because they're folding more airballs on the flop doesn't mean you don't wanna OB turn. If their range is still condensed then you still have an incentive to overbet.
Perhaps you're worried they're slowplaying too many traps? But just because they don't raise as often doesn't necessarily mean their calling range is more nutted. For example, say someone is only raising value hands and no bluffs. Yeah, they're not raising enough, but that doesn't make their calls stronger. (I'm not saying this is what's actually happening, it's just a common pitfall of MDA)
Something else to note is that your exploitative assumptions may be outdated.
Keep in mind there's a lot of bad MDA out there caused by simple biases. A really common one is showdown bias. Basically people assumed that every reg everywhere is underbluffing because they aren't accounting for this bias.
For example, here's the equity breakdown of BB after they check-raise on K72r. These stats are generated from GTO Solver hand histories.
55% Weak
And here's the same breakdown after I filter for hands that go to showdown:
31% Weak
So if you go check Stars HH or something, you'd assume that every reg is a value heavy nit. But the simple fact is that strong hands go to showdown more often, and that's all you see in your hand sample (unless you're playing iggy or something)
In my particular database of mid-stakes hands, players ARE overfolding on this flop, but they aren't too value heavy when they check-raise.
I agree with Tombos.
It was a good exercise for you to type this all out, OP. Keep up the good work.
Tombos21, looks like you are comparing flop range to river range. Am I correct? If that is the case ofc river range will be much stronger, there are 5 cards on the board.
I would compere how strong XR range that got to SD was on the flop with "normal" XR range to see how big this bais is.
Tombos21, looks like you are comparing flop range to river range. Am I correct? If that is the case ofc river range will be much stronger, there are 5 cards on the board.
I would compere how strong XR range that got to SD was on the flop with "normal" XR range to see how big this bais is.
Yes you're right, my mistake. Here's the updated flop comparison with the WTS filter applied. "Weak hands" goes from 55% to 42%, so bias is still quite strong.
Thanks for the replies everyone!
Whether or not to slowplay is not so clear. Ultimately you want to know if more money goes in through the XR line or the XC line.
You can't be certain that XR < XC just because they overfold to a raise. Perhaps they overdefend vs turn/river barrel. Or perhaps they're too passive after you slow play.
Also don't forget there are extra incentives to raise vulnerable value hands for equity denial.
That makes sense - especially I think in the pool I play in I'm seeing a lot of passive lines on Turn/River so I'm probably not getting as much value from bluffs or from some of their thin value as I should if I slowplay.
Disagree - kind of. The decision to overbet is more about polarity than range vs range equity. Just because they're folding more airballs on the flop doesn't mean you don't wanna OB turn. If their range is still condensed then you still have an incentive to overbet.
Perhaps you're worried they're slowplaying too many traps? But just because they don't raise as often doesn't necessarily mean their calling range is more nutted. For example, say someone is only raising value hands and no bluffs. Ye
My concern was them slowplaying too many traps - I guess it's necessary to run a solver and see how much that would change BTN play if e.g. they're slowplaying an extra 50% of 2p+ relative to theory.
My game tree doesn't have as many sizes as GTOW since it would take hours to run, but on a more extreme end of only raising 2p and slowplaying sets I'm seeing a drop in the 175% sizing from 15.8% of the time in aggregate on the turn (after BB checks) to 8.6% of the time. Some of that drop looks like it went into the 125% sizing, but the biggest increase is in the amount of checking the solver is doing on the turn. There's also a negligble increase in other sizings (I had1/2 and 3/4 available for the sim I ran), so I guess solver wants to keep polarity but just has to be stronger with the value end and therefore drop some of its bluffs rather than wanting to use a smaller size.
As for MDA - I don't actually have ANY data, I play on wsop.com and AFAIK there's no way to get any data around how people play on there. I'm basically just trying to infer things from showdowns and game flow. If anyone has suggestions on better ways to approach this I'm open to suggestions (e.g. is there some data source on this site specifically? OR are metagames similar enough across platforms and therefore some database from stars/GG/whatever would be a reasonable baseline to use to make these kinds of assumptions?) Even if that cross-platform assumption is true I'd need to find a source for that if I actually wanted to do some MDA.
Cool
Its small sample but MP and TP get to SD way more then 2p+ and trash/Ax high region. Which makes sense, those hands often check turn can call a bet, and trash keeps betting or check/folds. This is for this specific board? Would you share if have data for aggregate of more boards?
Cool
Its small sample but MP and TP get to SD way more then 2p+ and trash/Ax high region. Which makes sense, those hands often check turn can call a bet, and trash keeps betting or check/folds. This is for this specific board? Would you share if have data for aggregate of more boards?
Yeah it was filtered for K72r. More specifically, the above data is for: 100bb, NL200 rake, BTN vs BB SRP on K72r facing a 1/4 pot bet
I can expand the filters to include all boards in this formation (flop subset ~130 boards).
Facing a 1/4 pot c-bet, BB's response is as follows:
(Note that these numbers are very sensitive to bet sizing & rake structure. There's a huge difference between how it responds to 1/3 pot vs 1/4 pot)
Here's the XR range distribution: 62% weak hands
And that same distribution when filtering for hands that went to showdown: 57% Weak hands
Another interesting approach is to examine how often different hands take different actions. Facing this small bet, if you have two pair+ you're supposed to slow play it half the time lol.
You can also see that top pair type hands see a lot more showdowns compared to the trash in your range.
This last technique in particular is a nice way to build a strategy. Once you have the GTO hand histories you can break it down by texture and sizing, and get some relatively balanced heuristics without too much effort.
Cool ty.
So it's stronger but you mostly lose trash hands to showdown bais.
You often you hit 2p on very connected or mono board so they are not worth much. On JT9 BB maybe even the highest number of 2p, but they are far from nutted hands.
Interesting that middle pair and GS are losing hands. Guess it's defined as ev difference between facing and not facing a bet.
