matlittle

Thanks again Frankie Carson for offering to do this, it's been a really cool experience and I have certainly found some big holes in my approach.

For sure I have been relying on my RNG to make decisions rather than looking for good reasons to make a solid decision one way or the other. The KQ blunder vs UTG really highlighted my lack of study in tighter formation SRPs, which I have started drilling on GTOWizard.

I also drilled some OOP river spots with 2 bet sizings so I can get into the habit of working out the main value regions and find appropriate bet sizes easier. OOP river spots can be complex and you were right to point out my lack of clarity there.

I also need to spend lots more time on MDA research and accompanying nodelocking to work out which of the assumptions I have are actually found in the data rather than rely on loosely formulated exploits.

Lots to work on and I really appreciate the different perspective on my game which has highlighted all of this for me!

Fully agree with you that this bet sizing is slightly inappropriate and probably indicates that SB is making big mistakes with their range here. I think the main exploit here though is to attack the checking range rather than an inappropriately large cbet. Also, I would note the sizing and attack future block cbets in these scenarios.

I'm not sure I like raising the flop with 66 though vs this size and on this board type. The bigger cbet sizings are very poorly balanced (especially for OOP cbets). Plus, a reg should know this board lends itself to a block cbet sizing, so when I see this 60% cbet it just seems super strong to me.

This was an interesting hand for me too. I know Ad8 is a bet here at some frequency on flop. Looking at the sim it's almost a pure cbet, so I guess I was not cbetting it often enough here.

On the river, I thought this turn/river bet sizing combination would be significantly under-bluffed by regs. Looking at the data, I'm a bit off on that. My data shows a number less than the optimal bluff frequency but the combination of the difference and my sample size are not big enough for it to be statistically significant. I was also thinking the river being connected to the turn draws would make the river under-bluffed on average too, but that would require a lot of H2N research to confirm or disprove.

Another important factor is that my hand is not 0EV to call the river, it's slightly +EV.

So in conclusion I'm making exploitative folds based on assumptions rather than knowledge and I need to fill in these gaps with data research rather than relying on instinct.

Can I also please ask what your experience of playing these massive flop cbet sizings (250%) has been like? How well do people respond to them? Are there any exploits you think arise from using such a big size?

Q65hhd4d
Q65hhdKs
K87r4
K87rA
Here are a few boards I put in the solver and there is ~20% checking frequency on turn with 2 pair. If I change the turn cards for complete bricks, then it goes to ~0%. So I guess the heuristic is just that on brick turns you always bet 2 pair as its the nuts, but when the turn brings BB some hands that now are ahead of flop-2-pair you start to check at some frequency.

That's interesting, so on the T62r board there are more runouts where SB's overpairs can play for stacks, so it is happier building a massive pot on the flop on that board. Whereas on the JT3s board its way more dynamic so overpairs are worth less by the river, hence the desire to control the size of the pot a little more on flop.

I looked this hand up to see whether there is any turn checking with 2 pair hands after you cbet the flop for 27%. The solver never checks 2 pair there. On other boards though I know the solver checks some 2 pair hands. What is it in particular on this board that means it's 0% frequency? The turn being a brick? Is the small flop cbet size a factor too?

On this board you said you are playing 250% pot or check on the flop:

But then here you are using the 27% cbet sizing:

I am interested to know why you are going for completely different approaches on these 2 different board textures? I would have thought that whatever logic you used to the T62 board would apply to the JT3 board too, but I could be mistaken, so I wondered what the reasoning behind each approach was?

15:51 96hh i like a bet, bet, give up there with diamonds and heart suits. diamond cause obv flush drawl, and hearts cause spade and club combos fold turn. yea i get giving up river 9 high sucks, and dont think im pure barreling 96hh turn either, but do perfer over 96cc or 96ss

Yeah, close to my thoughts. I went 50/50 ott and you go pure. I assume you also BXB if you somehow miss turn

I was interested to see whether you get to bet some of these low equity 2-heart hands on the turn. For this 70% turn cbet size that you input, the solver is not really betting this hand class:

Instead it finds the occasional lower 2-heart hand, like 76hh or 75hh. Presumably having the J, T or 9 blocks more of IP's folds, whereas these lower ones don't.

This was a 3BP, and you said you would mostly raise this hand, I think you said 66% frequency of raising. I was surprised to see this hand class raising, but GTOWizard is doing it, so, good find from you. The frequency is max ~20% for these low JXs hands though. So did you over-estimate the frequency, or you felt like villain would over-fold to raises here for some reason?

Would have thought this would be a standard squeeze against a rec with a 13bb stack.

Also what do you think of the idea of raising to 7.5bb here, so the you get the chance to re-raise if the rec shoves? Really handcuffs the MP player and also saves you money when
MP 4bets you

Whats the thinking here of probing this hand in a 4-way pot? Looking at a 3way sim for a similar spot (CO/BTN/SB) the 77 is one of only 5 hands in your range that pure checks the turn.

Also, in these multiway spots how many sizings are you using on the turn? Are you just using 40% here or would you mix in something smaller or bigger alongside that?

I have a few tags for regs, this player is a weaker reg IMO (sorry Ivan). The broken stack and line choice would seem to back that up. Perhaps against this player type and with the river timing on the check I could just bet here. In game I was thinking QT would be the threshold for value but with these additional factors maybe I should bet the T9 too.

FYI this hand was vs a reg not a rec. They probed 50% on turn 3way, I called, then they checked river. Vs a rec I am definitely betting, but vs a reg I thought about it but decided against it. I felt the turn sizing was perhaps face up as TX, 99, maybe 87s or even 77 but then on river it's hard to get value given that I don't beat any TX the SB flats there plus I block 99.

Also in my mind was the idea that a reg here might think "I have lots of boats here I will bluff raise my 99/87s/77" rather than opt to bluff-catch those hands. All those hands block straights or boats and might look really appealing to raise given how many strong hands they could have there too, especially if I bet a milky looking 50% pot which I think is the max my hand could be worth there.

I was assuming Heat would balance strong in small.

KQ is the primary strong hand that the solver uses to balance the small sizing here though. KK prefers to bet bigger apparently, presumably blocking more value raises than KQ

This hand was a very player-specific attempted exploit. This player is too wide preflop in this scenario, very passive postflop and has very weak checking ranges. Even after node-locking those factors into a sim though, the only 2 hand classes really interested in checking here still are strong A high and really low pairs. So yes, you are right, I did push it a bit too far here.

The solver is almost pure blocking the KQ there. Looking at other parts of the game tree I cannot find a specific perceived leak that I would expect my opponent to have here to make me want to do something different. The solver (as IP) will:
1. Trap 14% of 2 pair+ on flop - would expect this player to find some flop traps for sure
2. Be indifferent between calling/jamming KX on river - villain shoved the AK here
3. Find bluff raises with hands like 76s, 66 etc - would expect to see that a fair amount although perhaps slightly less than the solver

Personally I think that would be too face up against a good 500nl reg. The block range would be vulnerable to bluff shoves and this player is good enough and aggressive enough to punish me if I never add traps to smaller bet sizes.

Hope to see you back here with a new contract soon!

In this hand, villain (50+VPIP) cold called a 3bet, probed 1/3 pot on turn and then checked river. Could you consider value betting the river here with the QQ after this probe sizing and river check?

Ah right, got it thanks. I think the squares being full threw me off completely. Plus I hadn't realised your small raise size. Looks like BB is required to 3bet almost every hand at some frequency which is funky

Thats interesting, do they not leave themselves really exposed to delayed cbets though after showing down this hand? If someone showed this down vs me I would DCB vs them whenever they checked to me

My thinking in this hand was that I'm probably supposed to call here with AK for a tiny amount of EV according to a solver, but then in reality most regs will just bluff almost always on brick rivers if they can't beat AK. I wasn't planning on calling river, so then the proposition of calling turn seems really -EV on that basis. Your idea of hero calling brick rivers sounds interesting though, there is certainly scope for SB to way over-bluff here when I look through the sim!

The reason I went for check-fold here was in part because the rec made it 4x preflop. It's also a rec that is mixing preflop sizings rather than JUST using 4x opens. I know from research that this range is going to be a fair bit stronger than if they had made it 2x or 3x etc. So I figured I have reduced fold equity on flop vs a cbet with a hand that is already supposed to mostly check-fold to start with.

I like what you were saying about the rec making turn or river mistakes though. I checked the stats for recs and they will under-barrel turn. This will get us more frequently to profitable river spots than we should do, where the rec has checked back turn with a really weak range like you mentioned. Will look for this in similar situations for sure!

One thing that makes it less bad is something you touched on at the end of your analysis Frankie. If villain is going too thin for value with this overbet sizing, e.g. with sets, then that adds some EV back to the play. This exact player made a play earlier on in the session in another SRP UTG vs BB spot that made me think this would be the case. If you remember this hand from video 1, villain made a pot sized river bet here with K6.

I remember seeing the showdown and feeling like the hand was too thin for this sizing, which the solver agrees with. The solver actually checks K6cc which villain showed down, and K6ss prefers the 50% pot sizing on the river here. So I already mentally noted this player was going 1 or 2 betsizes too big on river for value vs me, either because of a poor understanding of value thresholds or because they felt like I was a weaker player in some way.

I nodelocked IP to bet 99, 88 and 66 for this 150% sizing, and the KcQ gets close to 0EV as a bluff shove, assuming IP calls the shove with flushes and folds all sets. That's a bunch of assumptions to get it to 0EV though, and under all other circumstances its a torch, so even against this exact player I still don't like the shove at all.

Yeh this was the most interesting hand of the lot for me, because Frankie found a big blunder that I didn't even realise I had made. I made a bunch of assumptions that were wrong:

That I don't have many/any other natural bluff shoves as I couldn't work out any AcX hands that weren't flushes (as Frankie mentioned Ac8 and AcK would get here)

That I could shove K high flushes here as BB and so the Kc blocker was good

That UTG would have some lower flushes that most people would fold to shove (the board actually blocks almost all lower flushes).

This all points to a clear hole in my game in these tighter range scenarios. I spent lots of time studying BTN vs BB which is of course a good idea as its the most common scenario. Most video content is based on this idea too. But then my UTG vs BB knowledge is very poor as a consequence and I will spend some time studying this as a priority!

Also another thing to mention is that SB is supposed to donk here on the turn ~2/3 of the time. I think regs are getting better at recognising these scenarios these days, so I would guess that this player is finding donks there. If so, my TT gets to barrel turn almost always. But if they don't have a donking region there then I am supposed to range-check turn.

This felt like an awful spot especially with Ts blocker. I agree with the idea that SB will shove some flush draws here. My guess would be that most people are folding a decent chunk of KQss, KJss, KTss, QJss, JTss on flop though also, unlike the solver. Those are the main flush draw candidates to shove turn with, as hands like AJss, and ATss are more natural turn calls with some showdown value.

I agree with your suggestion Frankie that some players will just put the money in too often with 76s. Another hand I thought would be mostly shoved by humans was 99 here, can't see many people calling that again. That alone adds a lot of EV to my call so I don't mind it.

~23.10 is where you bring up the sim

Here we saw Clickr probe 75% pot with the J6cc, which to me seems like a really odd play. There should barely be any probes here from BB and JX (even with clubs) should just check the turn. Any idea what's going on here?

This was a pretty wild hand. It was 3BP, BTN stabbed 15% pot with AQ and BB XRed with T9hh. Any idea why BTN decided to size down to such a small flop stab sizing? The solver likes something between 40-50% pot and a polar stabbing range, very different to the 15% sizing chosen. It baited a low frequency XR and a 0% frequency turn barrel, so it worked out well in this instance.