Toy Gaming (part 2)
Posted by Sauce123
Video Player is loading.
This is a modal window.
The media could not be loaded, either because the server or network failed or because the format is not supported.
Resume Video
Start from Beginning
Watch Video
10
Loading 66 Comments...
Happy new Year :-) Looking forward to see this
good video,
For me at least, it really reminded me that having at least 2 bet sizings on the river will probably be more optimal than one in many situations. We were kinda discussing this in your last video w/ Adreno a bit.
Do you think it is fair to say that a good process would be (assuming your range isn't too merged):
1) Find a portion of your range you can bet as much as possible with and then balance that...
2) Take the rest and assemble a range for a lesser sizing.
Or do you think that such spots are definitely to infrequent to warrant that kind of regular analysis...perhaps because we are rarely polarized enough relative to our opponent.
-The better you can get at understanding how both ranges interact with the board, the easier this gets. From watching Ben's "stream of consciousness" style live video versus Syous, he seems to be able to regularly explore whether or not he can roll with one or two sizings on a given river (and a few flop) spots.
Noting that he said he made that video after a day of playing, I'm sure it's possible to do more than this.
-I've gathered (from various resources that try to apply GTO to HUNL, and a few small sims I've done) that past 2-3 sizes Strat-versus-Strat EV tends not to increase that much. Assuming three main sizings: underbet blocker types (5%-35% pot, like Sauce's turn leads HU when the board pairs), "standard 55%-90%" and "large 95%-infinity", the difference between trying to play a river range at optimal in realistic HUNL river spots with 1 and 2 sizings can be huge (3-12bb/100?), from 2-3 sizings maybe low single digits in bb/100 or less (fractions of a bb per hand).
-None of this takes into account huge Strat-versus-Strat swings in EV that can occur when your opponents play in a very exploitable manner, so for example versus fish trying to get your huge value sizing as thin as possible (and tracking the meta changes that go with it) might override my second point.
-Re: "assuming your range isn't too merged": I don't think it's that important your range be as polar as possible, you just need to break off the part you're going to play for the huge betsize, and rearrange the rest of your range so it plays reasonably well without the nutted part (of course you lose some of the bottom too).
I find in practice, ordering my blockers properly is kind of complicated. I'm not quite good enough to do anything more than the method you outlined.
sickmuse,
Good post. I mean "polarized" relative to your opponents calling range and "merged" relative to your opponents calling range (bc of course that is what matters sorry I should have been more clear). But you made a lot of great points in this thread and offered up some good research.
On another note, I always have trouble finding a balanced way to increase my EV with a smaller sizing (5-35% pot)...i guess if its super thin value where you have the right amount of bluffs to balance and you are often not going to get check raised...a very rare spot. In the spot you mention (turn leads OP when middle card pairs) I think would call for a larger sizing. One often has plenty of three of a kind in that spot so why not bet bigger (like pot-ish) so we can add more bluffs to our range?
I'll cover a couple of toy games dealing with this topic in part 3 or 4. I think you should generally focus more on putting your opponent in tough spots rather than explicitly thinking of balance in game. For me at least, when I try to put my opponent in tough spots, balance sort of "falls out" and I end up playing certain nut hands to various lines in order to induce various things. I mentioned the J8 hand because it's a clear spot for a huge overbet, because we're unlikely to induce many x/r with smaller betsizes.
Daniel,
You tend to say things like "One often has plenty of three of a kind in that spot so why not bet
bigger (like pot-ish) so we can add more bluffs to our range?" (my emphasis) Adding bluffs to our range doesn't really do anything for us (ev wise), since by definition bluffing with most hands has a near 0 EV. We add bluffs to our range so that a clairvoyant opponent pays off our value bets. Or we add bluffs to our range in some proportion to make us less exploitable versus smart human opponents whom we aren't confident how to adjust against.
Ben,
"You tend to say things like "One often has plenty of three of a kind in that spot so why not bet
bigger (like pot-ish) so we can add more bluffs to our range?" (my emphasis) "
Right, I realize bluffs are 0 EV against an opponent who will play us optimally, I'm not sure why I said it like that. The question I was trying to ask really is trying to think of a situation where a 1/3 pot bet with a normal PSR would make sense when we are representing the nuts? I never really use that sizing unless we are getting shorter stacked in the middle of the hand or I am exploiting.
"Adding bluffs to our range doesn't really do anything for us (ev wise), since by definition bluffing with most hands has a near 0 EV. We add bluffs to our range so that a clairvoyant opponent pays off our value bets. Or we add bluffs to our range in some proportion to make us less exploitable versus smart human opponents whom we aren't confident how to adjust against."
It increases our EV from our value hands though right, because we're betting bigger with our range?
Im really intrested as Daniel on this discussion about using sometimes 2 betsizes on the river.So you propose that on spots where we catch some backdoors nuts that we certainly know that our opponent can not have that hand of group of hands that we have, then create 2 betsizes ?
I was thinking on that, and i was worry to start doing for some reasons:
a) im affraid to allow opponents raise my river bets more happy if i capped my range by picking my "weaker" betsizing. I must to think that if we divide our range into 2 betsizes, one of the two ranges becomes more exploitable than the other one, since with one of those we have decisions when face a raise, and with the other one we have the nuts or just a bluff.
b) i was worry about giving information about the strenght of my hand.
You say that on those spots where villain range is very capped, those considerations are not relevant enough ?
Yes. I think in part 3 or 4 I'll solve some situations that will make your intuition more precise.
what if in the J8 hand turn goes check check and river is checked to you?
Good question. I'm not sure about this, but my standard would be to bet relatively small, and 3bet over a x/r. Usually in that spot OOP's range is capped at a weak-medium ace (and some people bet any ace), and OOP's range also contains lots of aces up combos and a few sets, as well as a lot of midpair type bluffcatchers. Against a range like that I'll maximize my value by bet/3betting, since I want to get some calls from the weak bluffcatchers, and the 2pair+ portion will x/r for me. The only part I might want to overbet against would be the medium aces, but those aren't usually too frequent in today's metagame.
Regarding your response to Privko's J8 question (on the board A749T). I understand why betting J8 smallish makes sense because of the ev gained when villian c/r you. It's also nice to have a few nutty hands in our smaller betsize range because it protects our thinner value bet from c/r's. I'm curious if you agree with my thoughts on how we might play some other parts of our range when we take a bet/check/bet line on this board.
If we had a weak Ax, I assume we should just bet 3/4 pot or so because your hand isn't strong enough to overbet. You'd balance this by throwing in some bluffs that that block his weak Ax hands (probably 12 or so air hands with a 2,3,5, or 6)
If we'd runner runnered 9T (for 2pr) I assume this becomes an ideal overbet candidate? As before, we have a polarised river betting, we are strong enough to overbet, and 9T blocks some of his 2pr hands that could c/r us. It also doesnt block his single pair Ax hands. We could then overbet bluff some more air hands that block his low Ax hands, or even some combos of a hand like 45o which blocks a set/2pr/Ax hands.
Does what I'm saying seem like a reasonable way to construct our ranges in your opinion?
I like your line of analysis. I think the missing piece (if any) is that implicit in your analysis is that OOP won't play clairvoyantly versus IP's overbets. Given an overbetting range from IP of [2pair, air] OOP's MES is to x/shove ~[T2P, some balanced freq of cards with good removal effects, probably Ax]. My guess is that an MES like this will incentivize IP to play the nuts for an overbet, which means that the strategy pair isn't at equilibrium.
Sauce blows minds. Where the hell do you come up with this shit man? props
Sauce are you saying equalibrium is where our opponent doesnt triple barell 1010+ here? Thats the only way I see this hand being applicable, unless you were planning on check folding the river after 3 bet bet bet with KK? I dont think people play this deep yet =)
Theory speaking, would it be a mistake to 3 bet KK here preflop? If you were 10 buy ins deep?
Ben,
You cover 2 examples in this video: a polar river situation (to introduce alpha) and a nuts-versus-air situation (to show how in a polar-versus-bluffcatchers situation, the polar player has incentive to increase his betsizing to infinity).
My question is simple:
What advice would you give to someone who feels like he understands these and more advanced concepts well but has trouble applying them at the table?
I followed along with ease during this video and others, but when I watch your other videos, specifically your live play two-parter, I notice how much slower I am in game, (1) estimating ranges, (2) applying reads so they're usable, and (3) applying the right concept to come to a decision about what to do with my hand.
Thanks,
Great question. Thanks for joining the discussion.
I'll get back to you on this in a couple days, but I've heard similar things from a few people who I talk with who have very good math/logic/analytical skills. The short answer is that people like you need to spend more time developing the skills you're naturally weaker in: stuff that somewhat fuzzily falls under "hand-reading."
Hey Ben,
I suppose I'll add that the way I think I should proceed moving forward, is to develop the skills you mentioned by playing sessions that isolate the use of that skill. I'd like to split off a session or two at the end of each day and play fewer tables, perhaps at lower stakes, and (1) get better at recalling/applying range estimates, and (2) find situations I can mark for further study so I have fewer spots where I must analyze something in-game.
S1ckmuse,
I think in order to play great poker it's necessary to make decisions at an instinctive level. I'm not sure if instinctive is the right word, so I'll offer a couple of other descriptions to make myself clearer. Psychologists (at least in the popular Kahneman book which I've read recently) sometimes split up our cognitive skills into System 1 and System 2, with System 1 being the characteristically automatic and effortless System that we use to read, see, recognize faces, etc. Athletes might say you have to "do it unconsciously" or "know it in your bones" or something. Programmers sometimes call fine grained instinctive decision making "expert knowledge" and they have trouble recreating it. I've sometimes said I try to think about the hand geometrically or spatially, by which I mean my mental representation is more like a picture than like a math problem. I'm sure there's loads of other ways to get at the same point.
My experience has been that what makes a great poker player is an instinct for making good poker decisions. In my own case I've found that directed study can do wonders for clarifying my thoughts about poker, which leads to training better instincts at the table. It's definitely true though that most pros (even high stakes pros) do very little study, and just playing tons of hands while thinking really hard about decisions trains them to play at a very high level. On the other hand, there's a lot of people who have played a lot of poker, and most of them haven't succeeded in becoming great players; I guess beyond a certain level some amount of talent is required. My view though is that dedicated study can go a long way, and it's a lot less of a gamble then playing tons of hands and hoping to be talented.
One training idea I've been working with lately is breaking up clusters of skills required at the poker table into categories which I label 'mechanics'. The idea is that by focusing on the conscious practice and refinement to instinct of one mechanic at a time I'll be able to build up a hierarchy of increasingly complicated skills. So, to take a non poker example for clarification, take driving a car. Driving is a fairly complicated skill, but one basically everybody can get good enough at. When we first drive, we have to attend to everything: the position of the pedals under us, the place the key goes in the ignition, all the various buttons, where the gas tank is, and then all of the driving decisions as well. But after a little bit, the basic mechanic I just described becomes automatic, and it's easier to focus on driving decisions; eventually driving is automatic (in most situations) and we can chat and think of other things while we drive. Sports analogies are good here too.
I think there's a useful analogy with poker. Basic mechanics are things like knowing the rules, knowing whose turn it is to act, knowing who is in what position, and knowing how much is being bet and what's in the pot. Intermediate mechanics are things like knowing what pot odds we're getting and giving, and roughly how good our hand is, and making some guesses about what our opponent is holding. Advanced mechanics might include things we're talking about here- instinctually balancing betting and bluffing ratios for various sizings, keeping track of our entire distribution at a given decision point, thinking one or more streets ahead in the game tree, and gathering probabilistic reads on our opponent to make exploitative responses to their set of strategies.
I find I learn efficiently when I get regular feedback on one or more mechanics I'm working on. So, you might go about translating your conscious, analytical, effortful work on poker theory into an instinctual mechanic by taking five or so hands a day and first writing down in a stream of consciousness your best guess at your range in a given spot. Then you might solve the spots to near equilibrium and see how close or far away you are. I find when I do this repeatedly my instincts improve.
This is all very sketchy advice, and I absolutely don't consider myself an expert (I'm barely a dilettante) in psychology or education or any of the relevant disciplines pertaining to this discussion. I have however spent a ton of time trying to teach myself poker, and hopefully some of the insights I've come to in my own study translate, or are at least food for thought, in helping you to turn theory into practice.
In a situation like the J8 hand how are you choosing which combos to bluff the river with? I would assume that most of the time we'd want to block some combos of the nuts but given that we don't think that villain can ever get to the river with J8 are we just looking to block set and two pair combos? If the answer is yes then isn't it hard to find many of those hands given our flop betting range or are you likely to start barreling off with 4x hands due to its blocker effects later on in the hand?
I think the approach you're describing is a bit too precise for a spot like this. It's true that if we could solve for the whole game tree (say from the flop onwards) we might choose some hands like 4x for our flop cbetting range in order to have better blockers on some runouts. But to make that decision, we have to weight the added blocker value of our hand multiplied by its frequency in the game tree relative to the hand's value taking any other line or combination of lines. And that's not feasible for us humans. So, I think the way to go is to sacrifice precision for some human level of balance, and just take roughly the right frequency of hands to bluff with. Like I said in the video, I try to do this geometrically. I start by noting that I'm betting 2x pot, so I need around 2/3 of a bluff for every vbet. Then I'll decide on some overbetting range (for simplicity assume it's exactly J8 suited)- that's four combos. Then I'll add some other suited combo from a similar part of my range to balance with, maybe 65s. I'll then take two or three combos of 65s and bluff.
The idea is that in this case the better our blockers are for bluffing the river, the better our hand is for checking on an earlier street and realizing more of its equity against a weaker range from OOP. When the pros for bluffing are couter-balanced by cons, I just try to arrive at a reasonable ratio as opposed to reverse engineering a flop range to bluff perfectly on an unlikely subset of runouts.
I was scrolling down looking for this question (it's an old video and I knew Ben wouldn't keep answering these comments forever).
I think it's pretty easy for some players to bluff not enough in spots where they can easily find the value hands to jam with, but have to try harder to come up with a balanced amount of bluffs.
(Maybe some players will overadjust by bluffing too much of course.)
It's easy to come up with the correct amount of bluffcombo's in hindsight, and pick a hand to do it with (like some combo's of 65s). But figuring that out in game is not.
Ben,
Thanks for your video!
You went through CREV analysis of the last hand (J8s) pretty fast and I wasn't able to grasp the value part of your overbet range. I'm not used to CREV because I do not use Windows so rewatching that spot did not do much good. Is it that narrow, comprised entirely of nuts and 2nd nut straights?
If I understand you correctly, you would split your river triple barrelling range in the J8s hand. This might be slightly off topic since you didn't talk about card removal directly, but let's just say that a villain knows that you're splitting your range on the river with 2 different sizings, and assumes that the only value part of your 2x shoving range is composed of both straights. So when a villain chooses a range of hands with which he could call your overbet shove, he should give priority to 87, A8 etc. over his sets, 2 pair or any other bluff-catcher, right?
Yes. But a clairvoyant player IP will realize this, and bluff with hands containing 8x as well. So then IP's bluffing range will succeed more often, and this will counterbalance OOP's choice of bluffcatchers. Another huge issue is that a clairvoyant IP can maximally exploit an OOP bluffcatching range of 8x hands for a ton of money if that range gets below a certain threshold for value. So, suppose OOP folds T2P (or even a set!) and bluffcatchers with 8x. That would maximally exploit a balanced shoving range of J8, 86 and bluffs from IP. But then IP just overshoves his own sets/2pairs, and bluffs with different combos and owns OOP's soul. The point being that there's a whole blocker-counterblocker exploitation game going on as well.
In one of the slides you mentioned delay cbetting and then betting the river being an example of when we're usually pretty polarised. Would you mind if I use this example to check that I have understood this video correctly?
When we arrive at the river like this our range is polarised and the villain's range (when he c/c the turn instead of probe betting) is mostly comprised of bluff catchers. Therefore if I understood the concept correctly this is likely a good spot to incorporate an overbetting range on the river. Correct?
Supposing we decide to use a river bet size of 2x pot. Villain must defend 33% to make us indifferent to bluffing. So when we construct a 2x pot overbetting range it should comprise of value hands that have >51% equity versus villain's 33% calling range and our overall betting range must be made up of 60% value and 40% bluffs
Would a good example of when to implement such a strategy be on a board such as ATx? The reason I think delay cbetting and incorporating a river overbet range here would be a good idea is because a lot of villians probe bet the turn with 90%+ of their Ax when we check back the flop. So when we delay cbet the turn and bet the river villain's range is mostly bluff catchers (Tx type hands) and our range largely polarised between Ax and air. Therefore this is a good spot to create an overbet range with the Ax hands we pot controlled on the flop and air hands.
Cheers!
Exactly right given your assumptions. If they fold more than 2/3 on the river, add a few more bluffs, and if they fold under 2/3 add some more value. I'd expect most players who fall under your turn assumptions (bet Ax, x/c Tx) to fold river more than 2/3 of the time. Also keep in mind that not all rivers are created equal- if the board pairs the "x" card, and that card is a trey or deuce, OOP will rarely have improved. But if the river card is a 9, every T9 combo hit 2pair and you're getting called.
Don't forget to hand-read! At least that's something I always tell myself when I'm applying a model to real poker.
I missed the answer on the first toy game. Early mornings suck.
Player A bets 1 to win 1, so he needs to succeed with his bet 50% of the time.
Player A bets pots, so player B gets 2:1 (33%) on a call.
But player B folds 2/3 times, so player A profits here.
Where did I go wrong? You say player B should call 50% of the time against a polarized range here, but he's getting 2:1 odds, not 1:1.
So does the answer lie in Player As bluffing frequency, and player B should call 50% anyway, at nash?
Player A should bet half the time, and be bluffing 1/3 times?
the betting range (A), if perfectly polarized, should have the same ratio of value:bluffs than the pot odds he offers his opponent (B); on 2:1 pot odds, the polarized betting range must have {67% value : 33 % bluffs} to make his opponent indifferent to calling or folding with his bluff catchers :
player B's EV is always = 0, whatever his calling frequency (be it 0%, 42,7% or 100%).
However, the BC range (B) must call at a frequency of 1 - the odds (as a %) the bettor gets on his bet; here, player A bets 1 to win 1 so needs to succeed 50%; Player B must call (1 - 50% = 50% ) to make the bettor indifferent to betting or checking : A's EV is always = 1 pot.
If player B calls less than 50%, A makes a profit on his bluffs; A can then increase his bluffing frequency.
If B calls more than 50%, A makes more profit on his value hands and may then increase his value frequency.
However, by diverging from the optimal frequencies, A runs the risk to become himself exploitable.
On 3:1 pot odds, A must have {75% value : 25% bluffs} to make B indifferent to calling or folding.
B must call (1 - 33.33 % = 66.66%) to make A indifferent to bluffing.
I cross checked several times this answer to make sure I didn't let slip in some mistake, but who knows ...
Edit : already 1 mistake corrected after posting, lol.
HTH.
If OOP only calls 33% of the time, can IP unilaterally increase his EV above the .75pot he should have at equilibrium?
I love it.
"It's a dominated strategy to ever bet with a bluff catcher [..]"
ofc, but why do I continuously leak money doing this ? :(
@Kokomo Thanks a lot, but that's exactly what I wrote.
The part I don't understand is that player A should win the pot 50% of the time when betting pot, and player B should win 1/3 times, given the pot odds.
But player B would fold 2/3 times, and not 1/2 times, thus giving player A a profitable bet with any holding.
I explained it differently because you are confused.
In a perfect PvBC river spot, when A bets 1 Pot, he has a ratio of {67% value : 33 % bluffs} so he wins 2 times and loses 1 time to B (EV A = 1 pot).
Because it's a perfectly polarized betting range, the bluff portion always loses to B's call.
So B's EV is always : 2 Pots [ Pot + Bet ] * 33.33% - 1 Pot (his call) * 66.66% = 0
A's EV is always : ( 2 Pots [ Pot + Call ] * 66.66% - 1 Pot [ Bet ] * 33.33%) * FreqCall + 1 Pot * FreqFold = 1 Pot.
Although simple, it's easy to get confused; I suggest you make a small spreadsheet template to play with the frequencies and bet sizes; add an EV table for A and B.
This will help you visualize the effects on the respective EVs of B calling too much or too little as well as A being balanced or not.
Disregard my previous response, looks like you already got there. You're right that if OOP only calls 33% versus pot, then IP can increase his EV by bluffing 100% of the time.
OOP has to call with a frequency so that IP is indifferent between bluffing and checking (remember, indifferent in this context means that the expected value of both options is equal). Since a pot sized bluff risks 1pot to win 1pot, it has to succeed half the time. So, OOP has to call with half his bluffcatchers to make IP indifferent with his bluffs.
The insight of the toy game is both (a) that making both players indifferent maximizes each player's EV versus a clairvoyant opponent, and (b) that bluffing and calling frequencies for 1 street polarized range games without raising can be calculated using the constant conventionally called alpha, which is equal to betsize/1+betsize.
@ 8:19 you say BB can fold KK. I don't know if I would fold KK there. Maybe I fold QQ or JJ.
I think you're forgetting that we're looking at toy games. OOP has a range of 100% bluffcatchers, all of which perform equally well against IP's range. The idea is that if IP bets only the nuts, and none of his air, then OOP should fold his bluffcatchers, which in this case were represented by KK.
Just look at it like this: You play with a three-card deck, containing a 0, 1 and 2.
BB holds the 1, therefore SB has to have 0 or 2. Nuts or air.
Thanks a lot, Ben.
I think the part where I got confused was the difference between BBs defense frequencies, and actually winning.
He has to call 1/2 times, but he only need to be correct 1/3 times.
And to follow that up:
Last example, Forhayley need to call 80 to win 120. That's 1.5:1 = 40%
So you should be bluffing 60% of the times here, and your play needs to work 2/3 times.
the bluffing frequency should be equal to the pot odds offered to call :
1.5:1 = 40% (= Min equity to call) => 1.5 value : 1 bluff => 60% Value : 40% Bluffs
... your play needs to work 2/3 times.correct; optimal calling frequency : 1 - 2/3 = 1/3 = 33.33%
J8 on A74 9T
You're overbetting your hand here in a spot where Forhayley is capped.
How do you split your value hands between all in and regular sizing?
I assume you don't overbet jam AQ here, so you'd only jam A7/A4/A9/AT sets and straights, but then your regular 75% or so bet sizing only contains thinner value hands, and Forhayley could check-jam over that bet.
Do you not jam hands that block his calling range (AA), but choose hands he can't have due to check-calling flop (like your J8, and 87s)? I guess that makes some sense because you don't need a wide overbetting range for value as you're bluffing 40%, but you have very few hands like that, and your two pair+ block a lot of combos that should only call a small bet.
We've talked about this in other places ITT and it isn't vitally important to the toy game to answer the question with precision. One example of a solution would be to overbet all nut combos which aren't present in OOP's range for an overbet (and add bluffs), and bet a different sizing with the rest of our range.
Bonjour ... i "see" a gap between your believe and your explanations ... you seems to praise highly the "visionary" ( seeing in terms of threeD of pictures landscape the big picture etc ) approach of the game but then your explanations seems to fall into the other world ( the analytical, math, logical etc ) like if it is impossible to "talk" about the game at the "visionary" level ... ? ... dont you think it is possible to find a way to teach to "see" ?
Hi Ben, i have sort of an abstract question.
Hypothetically speaking, if you had to pick a student (complete novice) to teach poker and have him beat lets say 5-10 for your uni final exam/thesis, how long do you think it would take you and what would be your approach of teaching trough - micro, low, mid, high stakes?
And if you had a choice to choose from 2 different characters which one would you pick/think would excel faster or be more challenging ?
1--Good at math, logical thinking but not very creative/imaginative and slow at deducting.
2--High level thinker, creative and imaginative, but sucks at math and lazy at studying on his own.
Why do you say that?
i think you should put "s+1" in brackets at min19
a story ... real ... i play checkers ( 100 squares .. much more complicate than the 64 squares one ) long ago and after one year and half i stop ( at a good french level, meaning pretty bad ) and for years u dont palt at all ... two years ago i play a little online and wanna understand more ... i find a young guy, world class level ready to teach me ... we to that over the phone playing online against others players ... after two weeks or so i tell him that i wanna stop cause he doesn t teach me anything ...he sound amaze saying " but i teach you everything i know" so i reconsider what he was telling me ... he keep on laughing, screaming, tell " look he s gonna play this piece, look that right side is so weak, the center is to tough, let him coming here .... etc .. he talk only about the "form" that take the two army, shoving the weakness or potential weakness here and there and play with almost only that criteria ......... i start to give more credit to what he s saying and start to think that way ............ and ........ few weeks after i start to draw agaisnt one world champion, and another, i do have a winning game agains rob clerc who was a great world champion .... i beat all the french players, kill them in fact ...................... if you know the game or similar ( chess go etc ) you wont believe that story ... when we work we never go over analytical trees like everybody does ... we never "calculate" unless a combination was available ... unless we have to ... this happen to me, the guy is "thomas m'bongo" .............................. to summarize you can teach me poker in two months Ben ... but you dont know how ........... and wont even try ..................... but ... this is a true story ... you are one person who can believe that story cause you dont use holdemanager for the right reason ... cause i always here a song believe the lines in your videos ... another story
cause i always hear a song between the lines in your videos
Hey Ben,
Hope Im not too late.
I was wondering how you would deal with card removal when trying to come up with a balanced river betting range.
In reality our opponent will have bluffcatchers in his range that block parts of our value range, some hands that dont block anything and he will also have hands that block parts of our bluffingrange.
How do we make sure that we are valuebetting and bluffing the right amount and not bluffing too much/too little?
loved the video btw. Would definitely like to see more of this kind with crev and stuff :)
Good question, but too complicated and off topic to get into here. This thread is for making sure people understand the toy games from the video. I'll definitely try to make a video on card removal at some point though.
Something specific would be using how using KQ on ATx22 blocks AQ AK, but the bluffcatchers range doesnt have AK or AQ to be calling with, and we'd say KT/QT are folding always so we want to block the AJ/A9/A8 etc? Hopefully you can chime in in part3/4, or at least how some softwares do or don't do card removal
Hi Ben,
You said that on the 633r flop in PLO betting significantly reduces ev compared to checking for OOP player after 3betting.
Does it mean he can't bluff air also? And what if he can bluff air could he balance it by betting some % of his SD value hands? Even though betting is -ev compared to checking combined with +ev bluffs (since he can't bluff catch air) it makes it +ev?
O,
Please stick to questions having to do with toy games.
Just finished redoing the CREV sims for both examples and I learned a lot by following the analysis then seeing how you translate that into the sims.
Very helpful to see how theory can be applied to these examples.
I find it difficult for me to use the different math equations, so I made a table with the main indifference frequencies related to the bet size; from there, I can instantly reckon all the others.
I thought others might have the same problem and find it useful, so I published it here : GTO simplified (OTR)
It was easier for me to commit this table and rules to memory than to memorize these abstract math equations.
Let me know what you think.Ben ... you say "I think in order to play great poker it's necessary to make decisions at an instinctive level" and after saying this you solve the toy game with a complete "calculations" solution .... my "story" ( true story and much more intersting that it might look ) was in fact of a way to ask you : is "instinct" not a different "approach" different "vision" of the problem ? ... is it another way to solve the problem ? this toy game in occurence ? ... i think those questions "stick" to the subject.
Based on your experience, assuming normal bet sizes, how much equity should be subtracted from a hand range when calling 3 bets OOP? For example, say the 2 bettor's range has 50% preflop equity vs the 3 bettor's range - in practical play how much equity do you think the 2 bettor should be given? Of course it would depend on stack sizes and positions, but would you say generally it's somewhere as high as 30%?
for example #1
in real game situation, our opponents will slowplay sometimes. how do we change our bluffing range base on that?
Seems like one has to be careful with sizing in polarized range betting. When we start to be large multiples of the pot over a finite sample the villain often regains almost 50% equity by always calling in the (AA,22):(KK) bluff catcher game. Hero is almost betting 50% each AA and 22 as the math permits almost an equal fraction of value bets and buffs to offer indifference to the villain – a clairvoyant villain, being indifferent, can now adopt any strategy and isn’t limited to an equilibrium choice.
The rigorous math still holds up because we check 22 once in a blue moon and win just extra often enough to make our expectation about 1 (the original pot size). But even though the villain is indifferent over an infinite sample they almost always do better by always calling vs always folding over a finite sample, and rarely much worse. Thus, practical considerations and actual sample sizes start to matter here. It appears this asymptotic result should be presented carefully. Things are clearer for pot size bets and the poker is also more sensible as fewer “human” considerations are involved when the bet size resembles the pot size. Variance is clearly less of an issue when bet sizes are similar to pot sizes as well.
Consider betting 99x pot=1, then Hero has 100(AA) and 99(22) in a betting range and one (22) in a checking range. The EV of this play is .99 pots: for example, the Villian is indifferent so they can fold every time, Hero wins (199/200) of the pot and Villian only wins 1/200 times when hero checks. Alternatively, if Villian calls every time Hero wins (100/200)*100 – (99/200)(99)= .995 - .05 (when we check 22 ) = .99 pots. They are the same as they must be and any other defense frequency, including the optimal 1% frequency is a weighted average of the same number giving the same EV of .99 pots.
The issue is that over a finite sample Hero is betting nearly equal amounts of bluffs and value hands and a strategy of always calling will very often capture nearly the full original range equity of 0.5. I believe this is a realistic issue that needs to be accounted for in a strategy. This is especially true for live play with limited opportunity to repeat the same scenario. This suggests the more common approach of building strategies around small multiples of pot size bets is generally more sensible but also that mathematically exact solutions and asymptotic conclusions should be treated with care.
Hi, I was just thinking because you mention the application of the toy game is mostly on the River. But I think often on the turn it's capped vs polarized and the capped player may still have decent equity but vs an overbet draws just can not call or raise (since capped). but maybe smaller bet and giving opponent a chance to make mistake with draws would be better. But I think it may be interesting to compare this EV with the EV you get from polarizing the turn.
I just signed up for RIO elite a few days ago and I just wanted to say: I read Janda's book Applications of no limit holdem a while back, and this video series so far and the comments section really put a lot of puzzle pieces together for me. Thanks!
Hi Ben,
I have been watching most of your videos recently. Great video. I guess one way to visualize the bluff catching situations is that the maximum achievable EV by the polarized player is never above the pot size (Bluff Catcher can have a simple folding strategy otherwise) , and since betting as big as possible insures bluff catcher folds at the highest frequency while having a 0 ev call it would mean betting max also maximizes EV by denying bluff catcher its share of the pot.
That was some heavy duty stuff. Great material. I'll start attempting some of this. But if I'm card dead, I'm exiting and coming back later. It's great to hear that so many great players at high stake are pushing this GTO paradigm to it's extreme. I can't wait to get to that level and try out some things.
Be the first to add a comment
You must upgrade your account to leave a comment.