Hi thank you for the video, very interesting, good format, very informative
AKs on Q42 w bdfd CO vs HJ 3b POT:
PIO vs PokerSnowie
...you demonstrate that PIO choses to 1/3 entire range here...
PokerSnowie choses to cbet 50% of range for 1/4 (1/3) but never choses AKs for a small bet and always checks...
I find it peculiar, that 'solvers' can come up with such vastly different 'judgements' of the same situation.
Do you find merit in developing a check behind range and putting AK (as the unpaired nuts) into it or no?
Why do you think it is, that they come up with such different solutions?
Aug. 18, 2019 | 3:30 p.m.
Dont you think most AKs willl be calls for him?! Mhh... welps
Getting in AK so deep in these spots feels very "meh" for both parties.... I don't really see 4b/calling AK as a big winner... unless we go with I dont know, ~50% of AKs and 10% of AKo and the rest are flats pre
What do you think about reraising small to 'force' some of his Ax to call/have a tough decision.
Aug. 17, 2019 | 4:10 p.m.
Interesting spots, good commentary
You "bash" the big 4b w AA and say its "too big".
Don't you think that is entirely dependent on how he designs his range? (Because technically the polarization and hand selection does in any spot define the 'possible' size of a raise/bet, right?)
What is the worst hand you expect him to 4b that size and call vs a shove?
I am hinting at the fact that you are probably 'only' folding out his Ax 4b bluffs and in the best case get called by KK, don't you think?
I learned something regarding how you change your thinking and strategy based on how you perceive/mark your opponents. Since I tend to always look for the "GTO" solution that is very refreshing for me to see.
Aug. 16, 2019 | 1:53 p.m.
Thx Felipe Boianovsky
I admit, the post reads a little bit too "riled up" and has a ranty exaggeration-flow to it.. haha
" I do think there are still a lot of things that we make the mistake of "humanizing the solver", but everything it does IS for a reason".
This is "true" and "wrong" IMO at the same time. I obviously see what you are getting at but saying "everything it does IS for a reason" can imply that these reasons are "different" in nature ... and see, THAT is exactly not the case
there only is and ever will be 1 !!! reason a solver does anything, and that is to increase EV - which is a number behind an = in an equation.
So there is an AI that has the "task" to maximize a number, given certain rules and circumstances. And it does that by trying out different lines and sizes with its range and combination however-many-gazillion times vs villain - and it 'simply' searches for the solution that has the highest number behind the = in the given formula.
By trial and error - not by "thinking" about what its opponent might do.
It doesnt know that when it bets with 9x on A9x x on the turn that his opponent folded Ax, - it cant "look" into the others strategy or understand it. It just knows that 9x, for whatever underlying mathematical reason, is better than, say 66. Most likely bc if called it has more outs and it gets called less ofen bc it blocks two pair and set combos.
So imo it is a "mistake" to look at a solver solution and say "oh so it is to fold out Ax" but its better to categorize the situation into a folder and then to apply the concept of "midpairs on A high, non-straighty, flushdrawed turns where my opponent is so-and-so capped to XX as his best hand make up very good/necessary non-FD, non two-pair or better big 2nd barrels"
Aug. 16, 2019 | 1:13 p.m.
Thank you for the video, the effort and the examples!
I find it funny/interesting, that people/you are still using this 'human' lingo when they talk about solver solutions...
Its very pervasive and so interesting to think about... because:
(as an example)
16.38 in the video onward: A93r 2c board you say something like "even when we bet 9x here it is as a bluff in order to make JJ, QQ and some Ax fold"...
The solver doesnt "think" and isnt doing something "to make someone fold"...
Also solvers are never "bluffing" - solvers are always "value"betting. Because when the EV of bet is higher than the EV of check, than it is a valuebet...
calling it a "bluff" is merely a human abstraction.
A solver is simply presenting you with an equilibrium... even more technically it isn't even ever 'value'betting because it is playing vs itself and therefore longterm the "true EV" is 0.
It therefore is just "value"betting to prevent itself from losing.
Its solution is simply a calculated solution to a given problem, that presents the counterpart with the dilemma that deviating from its own equilibrium will result in an overall smaller EV.
not to spill the tea here BUT
Geometric betsizing in all sorts of scenarios is right now IMO the biggest typical human leak pervasive in almost all limits..
when you study solvers and PC vs PC solutions a lot and deepy you will find that there are VERY VERY VERY few situations where a PC lets say
3bets in position and then goes bet bet bet (allin) in a geometric way..
the same for OOP and even more rare for single raised pots in and out of position (in terms of bet 3/4 three times)
These lines are (theoretically) reserved for VERY few VERY specific runouts, usually where the rivercard pairs a non-playing flop or turncard and does not change the range vs. range interaction at all.
The oldschool learned "bet bet shove" geometric 'paradigm' is a pureley human, usually highly exploitative approach, mostly heavily weighted towards 'too-much-value' and is/was useful vs human players with human leaks and cognitive biases in the old "we are the regulars and there are the fish"-days and will usually be at the expense of overall EV for heros range.
PC solutions have A SHIT TON of bet small bet small bet BIG and polarised or bet half bet half bet small, or bet BIG, bet small, bet half lines. These are obviously very hard to implement and easier solutions might therefore have higher actual human EV.
But my main point is that it is time to distance oneself from these terms like "betting to make him do this and that" - bc a good strategy can't, shouldn't and doesn't care what anyone does with, lets say JJ on A93r 2c turn. It works bc it taps into a deeper layer of mathematical truth that runs in the background of the game.
Aug. 8, 2019 | 5:56 p.m.
all the decisions so far.
Hope to get a beta invitation
1. fast forward or normal?
2. Given it is "normal" poker, given I join with 100bb, given I double up, given there are 20 tables running, given I got assigned Peter S. as my name, given I then quit with 200bb to "pocket the win", given I then rejoin 1 table...
am I still Peter S. and am I still playing on the table I left before that?2.1: When I join 4 tables, am I the same Alias on all 4 tables?
2.2. How many tables maximum?
- have you discussed/come up with
in which way you'll provide premade %-of-pot betting options!?
June 11, 2018 | 1:59 p.m.
One thing is for sure and that is you DEFINITELY² don't reraise preflop vs an unkonwn IP 4b from OOP 250BB deep with AKo
Thats very likely a huge blunder and money burner
As played I think flop is fine but the turnbet I feel is way too big and has no place considering the ranges that are playing against each other here..
You really just hope he has QQ and puts you on a picked up flushdraw on the turn
On the river your only hope is he turns QQ into a bluff or is a huge spot and now somehow bluffs a busted turned flushdraw...
maybe you won, who knows but I think pre is the real lesson here and the turn is too big too
Nov. 5, 2017 | 2:40 p.m.
So, Solver does bet here, but choses a smaller sizing and plays a mixed strategy of bet and check on this board with all its Jacks, Overpairs and Gutshots (QT) leaning towards betting everything that has a backdoorflushdraw or blocks the backdoornuts
After betting small on the flop it then wants to bet small again on the turn with
and checking only JJ and 88
Overall it's always gonna be a tricky spot when gettting raised anywhere in the hand.
Overall I feel the hand is played fine
Maybe there is something to be learned about the fact that the solvers rather "blockbet for information" (highly inaccurate statement) even with AJ and go from there and only "induce" w the absolute nuts
Nov. 5, 2017 | 2:19 p.m.
So a solver wants to definitely check your hand on the flop
from the turn onwards betting your size but maximally POT on the river is ok with ~20%ish frequency..
You definitely can't do this all the time "because you block the A hi NF"
Nov. 4, 2017 | 2:10 p.m.
Hey, I thought about two concepts for applying overbetting and "finding spots" recently and I would like to discuss them...
The first approach that I also feel is the one that any solver would take is sort of
- "overbetting when it is dictated by position, board structure and the subsequent strength of both players ranges"...
What you find for example in Pokersnowie is that on certain boards in a certain spot his recommended betsize for the entire range would be 2xPOT which it then does with a polarized range and then he checks a lot of the hands that any "normal regular" on typical stakes would usually bet 1/3 to 3/4.
- The 2nd approach, that I feel is most usually used by the "average reg" is the sort of approach where your own hand gives you
a) an information advantage and/or
b) a strong incentive to overbet the pot for certain
which CAN but usually IMO doesn't overlap with a general analysis of the board structure / equity distribution of the ranges.
So an example for 1 (in Pokersnowie) would be:
UTG open 1/2 POT BB flats
Flop: AK6 rainbow
Snowies proposed betsize is now 2x POT and he does it 100% with:
AA, KK, 66, AK, 55, QJo, QJs, QTs, JTs, A6s
and then >70% with: 77, 88
and checks: A2-AQ, any K, TT-QQ
whereas "general population" would probably
with any A, some Kx, Gutshotdraws etc.
A sort of typical example for 2 is lets say:
HERO Bu opens 33, BB calls
FLOP: T37 rainbow
check, bet 1/2 POT, call
check, bet 1/2 POT, call
check, Hero now "overbets because BB can never have TT, some draws bust, he can herocall with Tx and because he is greedy with quads" whereas a solver (Snowie again) plays 1/2 Pot river w Tx, 7x, 3x, JJ-AA and some bluffs simply because the range distribution dictates a 1/2 POT betsize.
Sometimes on the felt I am torn apart between these two concepts since in general I feeld concept 2 is largely overused by most players and therefore allows for a ton of betsize-based hand reading...
it is usually much harder to be alert and clairvoyant enough to always be spotting spots for hard polarization as seen in example 1 and then execute with the right amount of especially air (because the correct amount of air is usually not just the most obvious combodraw/blocker etc.)
But I can obviously also see the upside of simply looking at the information advantage of your own hand (e.g.: unbeatable) and then pound on a likely weakness of villain of calling too much bc of herocall tendency or whatever.
Soo... what is your approach? I at least feel I way too often use my cards to determine wether or not I want to overbet instead of the positions and the board texture.
Oct. 27, 2017 | 12:23 p.m.
UTG: $100.00 (Hero)
Oct. 24, 2017 | 11:40 a.m.
All depends on assumptions,...
maybe he is a sicko and is shoving K of club blockers ;)...
He could be "whatever" and shove AT, TT, 99, A6 that he for some reason "slowplayed"...
overall after we slowplay and super underrep our hand maybe we should just reraise on the turn
Oct. 22, 2017 | 10:10 p.m.
DEFinitely at least 1/3rd this flop
dont reverse lvl yourself,
there are spots to check topset (A72r after sqz vs 2 opponents etc.) but not this one
turn is fine but could be bigger I guess and the reason to reraise is probably actually to not kill your action vs him thinking that his 66 is no good anymore bc the flush/straight comes in etc.
His range will probably be 78, 66 and then combo/picked up flushdraws
I guess you post the hand bc you lost to a flush ;)...
and tbh there is a point bc with the flush coming he probably won't shove 66 anymore
The real mistake is the flop
Oct. 22, 2017 | 6:09 p.m.
Preflop can be a mix of reraising and calling
turn is actually more often a check for overall range "stabilization" (according to snowie)
bet is obv fine though
but then it's a crying call, not that +EV but probably > fold