matlittle
4227 points
Yeh I think it's the right heuristic but the wrong logic. I had assumed that AhA would have more equity (blocking strong draws) and be more resistant to heart rivers, so wanted to necessarily increase the pot by betting. Then over-pairs with no heart prefer to check-shove and end the hand now. Not sure that logic is true either, so I guess we are both making the right plays but neither of us knows exactly why!
Jan. 18, 2025 | 2:03 a.m.
I had a very similar experience last year, playing live poker for the first time in years. Just listening to the logic (or lack of) behind the decisions of recreational players was very revealing.
Jan. 18, 2025 | 1:56 a.m.
I have a fun example of exploiting loss aversion. I have a friend who I play regularly in a home game - whenever he is in profit he separates his chips into 2 piles: his original buy in and a 'profit pile'. Whenever he has a tough bluff-catching decision on the river he picks up his profit pile, counts it out, and sees whether he is still in profit if he calls and loses. When his profit pile is low, he is very easy to bluff, especially if you bet an amount larger than the profit pile. This effect becomes very extreme towards the end of the night, given that losing a big hand right before the end of the night might take his session from a win to a loss, and he will almost always opt to lock in a win instead of making any kind of bluff-catch at that point.
Jan. 17, 2025 | 2:57 a.m.
Really interesting video Sam! I have always wondered why recreational players sometimes make insane call-downs in big pots. I guess loss aversion and not wanting to take a loss is a big part, as is emotional attachment to the hand they have or the money they have already invested. Seems like some players, once they have invested a certain $ amount or percentage of their stack, won't fold the river, so bluffing them will cost you huge amounts.
Jan. 17, 2025 | 2:07 a.m.
These ranges in the bunny software are your own personal ranges that you have loaded into the software, or the software comes with them already loaded?
Jan. 16, 2025 | 1:30 a.m.
the grouping of boards like AQ7 and T22 in same bucket.
These 2 boards were presumably both cbetting mostly for the overbet sizing?
Jan. 15, 2025 | 10:37 p.m.
At this part of the tree the solver was playing mainly raise/fold with little calling. I think this was a by-product of removing the option for IP to 4bet the flop. Before you removed it I think it was 18% raise frequency, then after it is 41%. If you remove the option for IP to re-raise, and the flop 3bet sizing for OOP is very small, then OOP gets fold equity, maintains the initiative and gets to see the turn all for a very cheap price. I agree with your idea that real humans don't really 4bet IP in this spot though, so perhaps the min 3bet for OOP is in fact a decent play in reality too?
Jan. 15, 2025 | 3:11 a.m.
Cool video, its great to see these crazy hands under the microscope.
You mentioned here on this turn that AA and KK with a heart will bet more often than without, because they unblock the stabs. Do they unblock the stabs because we don't expect IP to bet too many flush draws at low SPRs because they don't want to bet-fold equity?
If they unblock stabs, then wouldn't it be better to check them, rather than bet? Presumably we want to induce bets with value hands?
Jan. 15, 2025 | 2:58 a.m.
There are many boards as we just saw where the raising fqcy is basically lower than 1% yet we still are getting raised lol.
I see this all the time in my games. I think there are 2 primary reasons for it:
1. People don't study enough to know that the raising frequency IP is low in general and close to 0 on certain boards.
2. Using sims with multiple cbet sizings leads to a higher raise frequency when facing the smaller cbet size(s) because the smaller size(s) are naturally a little weaker. So the sim says you should raise way more if someone bets small, but it's not correct if that person is in fact using just 1 cbet sizing. For example, using GTOWizard pre-fabricated sims or flop reports to study will cause you to raise too much here.
I think that a lot of players raise too much IP in 3bp w the wrong combos
Agree with the 'incorrect combos' part too. Which types of combos do you think people are incorrectly raising with here, based on what you have seen?
Jan. 15, 2025 | 12:29 a.m.
Great video! I think one of the reasons players miss 3bet bluffs on the flop in 3BP is because they feel they need to shove due to the low SPR. That leads them to only raise with decent equity hands - flush draws, strong pairs, pair + draw etc. By having a non-all-in raise size we can include more bluffs in the range.
Jan. 15, 2025 | 12:19 a.m.
Do you have any tips for cold 4betting vs recreational players? I assume that we just go super linear and forget about hands included for board coverage? Anything else important to consider?
Jan. 14, 2025 | 11:56 p.m.
I see that we managed to bully you into sitting in the bottom centre seat on the table! Or perhaps you just didn't change the seating settings on GG yet? How come you have decided to play these games instead of stars? How did you find them so far? Do you have a big enough sample to compare win-rates across sites?
Jan. 14, 2025 | 11:43 p.m.
Zoomed in on the image a bit:
Jan. 14, 2025 | 11:39 p.m.
In this scenario you were looking at UTG open, CO 3B, and then SB cold 4B. In the cold 4bet range for SB it doesn't have full frequency of AKo, but has small percentages of KQs, KJs, KTs, ATs etc. I understand that this is for board coverage, but do you think this solver range would perform better in real games than a range that includes 100% of AKo, then skips these marginal suited broadways?
Jan. 14, 2025 | 11:38 p.m.
Great topic to cover, and I'm glad that it's in this mini-course format. I think that some added structure to the video content here is definitely beneficial, so this is a great idea!
Jan. 14, 2025 | 11:31 p.m.
Hey Frankie, that's a very interesting way to cluster together different board textures! It seems like the algorithm did a pretty good job in grouping together the boards. Were there any big outliers, i.e. boards that didn't seem to fit in their group at all? Or just a bunch of borderline boards that can go into multiple categories? Did you do the K means clustering in Excel? Is this an in-built feature of Excel?
Jan. 14, 2025 | 10:32 p.m.
do you think this general theme holds true for 3b pots facing triple barrels as well?
+1 for a similar video on 3BPs (or any other similar river scenario)
Like you said Tyler Forrester, there is less scope to look into bet sizing, but I'm sure that there are important patterns across board textures when it comes to bluff frequency.
Jan. 9, 2025 | 6:18 p.m.
Too focused on equity driven strategies.
Based on this part of the data, you are not alone! Looks like most people miss the low/zero equity bluffs that can bluff rivers that complete high straight draws. That might not necessarily be a mistake though, if your opponents call down too light then it might be preferential to bluff only high equity hands.
Jan. 9, 2025 | 6:16 p.m.
Hello Tyler, really good video! I think these H2N insights are really helpful.
Can I ask whether these board textures on the left of the picture are all referring to the board texture by the river? Some specify that, but others don't, and I know you can get filters for flop textures too.
Looks like the general pattern amongst this data is that if the 2-broadway region connects well with the board, then river bluffing frequency is low, and if it doesn't then bluffing frequency is high. I always assumed some pattern like this would exist, but it appears to be more extreme than I had imagined, very cool to see the data like this!
Jan. 9, 2025 | 6:03 p.m.
Your opponent showing down AA here though probably means that many players are good enough now that you can't use this 2e sizing on turn.
Jan. 7, 2025 | 2:25 a.m.
This was a really interesting hand. I was dubious about if we would get to have a turn e sizing here. At equilibrium it is not possible, but if you nodelock a slightly lower frequency of QJ in IP's range, then we can use it.
For the river, the solver prefers a smaller sizing though after the board pairs. We will have boats here, but straights makes up a far bigger chunk of our value region and they can't shove and prefer betting less than pot.
Interestingly the hearts make for worse bluffs than the diamonds on average. The reason being that AX with a flush draw is a big part of IP's river folding range, and the Ad being on the board means they can only have AXhh for this hand class.
Jan. 7, 2025 | 2:20 a.m.
the board texture (3 broadways) is generally a board that is under raised compared to GTO
I didn't do any MDA study on board textures, but I was thinking the same thing too. Opponents probably wary of their lack of sets/2 pairs, plus large pairs making for seemingly good bluff-catchers probably affects the decision making process here of regs causing them to under-raise as a bluff. Plus of course your exact blockers here block the naked flush draw hands that can bluff raise
Jan. 7, 2025 | 1:21 a.m.
These days for the river spot, there is a subset of regs who use MDA based exploits who will go a little crazy with bluffs vs a river block bet. Against them it might be highest EV to block the river with nut combos, despite knowing your opponent is probably very capped, given the frequency at which they will bluff raise. In this exact scenario though with this combo that probably wouldn't work as your opponent is very likely to have and A or K that will just call, given your blockers.
Jan. 6, 2025 | 3:43 p.m.
Happy new year Tyler!
Enjoyed the debate around this hand. On turn, you were deciding between check, bet small and bet big. Are how many delayed cbet sizings will you play here? Are you utilising both a big and small bet?
You mentioned on the river that you will bet big because you don't expect many raises here as your opponents will not have flushes anywhere near enough for them to raise here often enough. If this player is a reg, can we not apply the same logic to the turn and bet big there too, given that MDA data shows that the average reg will under-raise the turn here, so we strictly prefer the bigger sizing with our nut combos?
Jan. 6, 2025 | 3:39 p.m.
Do you think this was an unforced blunder from your opponent? Or was it potentially some kind of exploit expecting you to call down too light? I would assume the former given how much EV you lose at equilibrium with this play.
Jan. 5, 2025 | 3:46 p.m.
I would also assume that most humans might bet a bit smaller with a set here on the 4 straight river, yet put most of their bluffs into the 3/4 sizing because they want more fold equity, so then this big sizing may only rep a 9 for some players, and is probably way over bluffed. Would you agree with that or not?
Jan. 5, 2025 | 3:34 p.m.
Hey Frankie, happy new year!
This flop is actually better for IP than you would think. MP doesn't open 33 at full frequency, then doesn't call all of 77 or 55 to the 3bet. None of the suited connectors flop straights. So then OOP isn't as strong as you would think on this board, can't lead, and IP can cbet a reasonable frequency.
Having said that I think your line is great. The spades are supposed to be good to bluff-catch the river too (against a good opponent).
Jan. 5, 2025 | 3:33 p.m.
Do you know whether there have been lots of hero calls in these games in general? I was thinking that there have been, but my opinion is probably skewed from having just watched 40 minutes of light call-downs.
Jan. 2, 2025 | 5:12 p.m.
Hi Callum, happy new year! Good content for review, these games have been interesting to watch. Look forward to the solver review, if you get more interesting hands in the future then I would be keen on another similar video. Would prefer the names to included personally, not sure if you can get that into the replayer or not?
Hello Gary, great video! Some really useful exploits in there, would be keen for more videos like this in the future for sure.
For this AK hand, I like the line you took and the logic behind it. Do you think this line performs better than checking the flop to either check-call (keeping the recs range even wider), or check-raise (expecting opponents to overplay all their 1-pair hands)?
Jan. 20, 2025 | 10:45 p.m.