HayStack

Frequencies in a Haystack

Through personal work, I’m convinced frequencies truly are king. I did some work tonight with PokerSnowie Coach. The goal was to find some patterns in frequencies so I could estimate the correct-at least ballpark-frequencies based on patters in the changing factors. Unfortunately, this goal seems perhaps unattainable. I’m hoping Snowie is wrong with these numbers. Here’s what Snowie told me in these two simple situations.

FrequencySample

This is incredibly discouraging to me. I’m not finding any pattern or rule of thumb remotely useful. The number of situations seems incredibly daunting; too many and too far apart to use rounding based on situations. The frequencies fly all over the place. Hopefully something is amiss and a method exists to ballpark scenarios.

Ugh.

Like this post? Receive free tips and articles from QTipPoker.

facebooktwittergoogle_plusredditpinterestlinkedinmailby feather

Books by Owen Gaines Forums Frequencies in a Haystack

This topic contains 30 replies, has 4 voices, and was last updated by  QTip 3 years, 5 months ago.

Viewing 15 posts - 1 through 15 (of 31 total)
  • Author
    Posts
  • #800

    QTip
    Keymaster

    Through personal work, I’m convinced frequencies truly are king. I did some work tonight with PokerSnowie Coach. The goal was to find some patterns in frequencies so I could estimate the correct-at least ballpark-frequencies based on patters in the changing factors. Unfortunately, this goal seems perhaps unattainable. I’m hoping Snowie is wrong with these numbers. Here’s what Snowie told me in these two simple situations.

    FrequencySample

    This is incredibly discouraging to me. I’m not finding any pattern or rule of thumb remotely useful. The number of situations seems incredibly daunting; too many and too far apart to use rounding based on situations. The frequencies fly all over the place. Hopefully something is amiss and a method exists to ballpark scenarios.

    Ugh.

    #802

    Sevendeuceo

    I’ll just focus on the 100bb effective stack size for now.
    Many of the results seem broadly consistent with Miller’s 60% to 70% ballpark figure (unadjusted for “bad events”). I am unsure how to interpret the bet size options though. Is it the case that Snowie would itself always use the recommended size with the corresponding frequency, and the frequency for other bet size is an alternate sub-optimal strategy?

    First BTN IP — BB flatted: Maybe the explanation why the frequencies are so low when the chosen size is pot is related to the concept “Don’t Take Away Their Rope” from Miller and Sklansky’s NLHETP. Betting pot might just be too heavy a price for Villain to call with much of her range and hence we are forcing her to play close to perfectly (in Sklansky’s sense) through folding with a very high frequency (and only calling with nutty hands). Hence, if we are restricted to using such a large bet size then we might need to polarize our range with almost only nuts and some draws to the nuts (nut-busters) hands? That might still be exploitable but just is the best we can do given that our bet size is too large. However Snowie does recommend the large bet size for draw heavy boards (except the high one — which may be because the BB range is draw-poor on high connected boards, since such draws would mostly consist in easily dominated hands).

    When button flats our CO open, then Miller defines this as a bad event and we are justified in reducing our betting frequency on that street. This is especially true when the board is draw heavy and/or low since dynamic boards favor the IP player. So, here, Snowie’s recommendations seem to qualitatively match Miller’s rule of thumbs. The highest recommended frequency (39-43%) are for high and/or high-dry boards that don’t penalize the OOP player nearly as much.

    The figures on the right hand side of the table are more difficult to interpret. I’ll come back to them another time. Those are my thoughts for now and some of them might be rather half-baked. I agree some of the results are counter-intuitive (and some may have a clearer rationale when looked at in the context of Snowie’s game plan for the later streets.

    #803

    Sevendeuceo

    I wonder also how the frequencies on the top left quadrant relate to the BB’s donking range. Snowie came up with those optimal frequencies through playing itself. It must have some donking range since it cares little about the etiquette of checking to the raiser. So, some of the lower listed Button betting frequencies (on mid drawy boards) may reflect the fact than on such boards the BB range may be narrower since when we are checked to we are facing the BB’s preflop range minus its donking range.

    #804

    Sevendeuceo

    I didn’t quite completely express my thought. The drawy mid-boards smack the BB’s range the hardest. The BB will more frequently donk (though with a balanced range) in order not to let the Button check back and get a free card. Hence when the BB doesn’t donk on such boards it’s more likely to be weak or on a draw. We thus have less incentive to bet since there isn’t as much value in doing so unless we ourselves need to protect a vulnerable made-hand.

    That may mean that the BB’s checking range on such boards is somewhat unbalanced. But that’s consistent with the overall BB’s strategy being GTO since it is permissible for a GTO player to partially “give up” on some branches of the decision tree. (As is the case, e.g., with our own OOP turn check range OOP as the pre-flop raiser, in which we only might want to include the odd booby-trap if any.)

    #805

    QTip
    Keymaster

    I wonder also how the frequencies on the top left quadrant relate to the BB’s donking range. Snowie came up with those optimal frequencies through playing itself. It must have some donking range since it cares little about the etiquette of checking to the raiser. So, some of the lower listed Button betting frequencies (on mid drawy boards) may reflect the fact than on such boards the BB range may be narrower since when we are checked to we are facing the BB’s preflop range minus its donking range.

    Sorry for the confusion. The top left is Button’s betting frequencies after BB checks the flop.

    It seems the biggest factor in determining the frequency of the cbet is our bet size. For most pot size bets, we don’t usually bet after preflop. Actually, the two hands I worked out on here (both open from UTG: one MP flatted and one BTN flatted), snowie checked about 100%

    I want to mess with the turn now to see what the frequencies look like there.

    I started messing with some bluff:value ratios as well. This chart took a while to build. The first frequency hand I did took forever. So…this is going to take some time to sort out.

    I’m not sure what to think about snowie’s quality yet…wish we knew more about the accuracy of its decisions. I also need to spend more time with janda’s book to see if I can learn what makes sense with these frequencies. I do know after reading most of the book, the 65%-75% frequencies came up often. But really, if you look over this chart, those numbers almost never occur, especially for continuation betting the size of the pot. This is even in position on a dry board.

    Open on the BTN, BB calls. Flop is J72r. BB checks. If you’re betting pot, the cbet frequency is 37%. Who saw that coming? A83r: 25% frequency at pot-size bet. What?

    Open in the CO and BTN flats. If you’re betting pot cbet J72r, the frequency is 5%.

    I’m not sure how categories of flops we’re dealing with here. Certainly we have like 19,600 flops; however, there is reduction with the suits having no importance. I don’t know what that takes different flop textures too, but the point is, it seems the frequency is more sensitive to texture than I originally thought. Then we have different bet sizes to use. Certainly we can pick one size and go with it. Even then dealing with just these two scenarios in real play is intimidating. We’ve got dozens more to examine. Perhaps they’re simpler or similar…idk yet.

    • This reply was modified 3 years, 6 months ago by  QTip.
    #809

    Sevendeuceo

    “Sorry for the confusion. The top left is Button’s betting frequencies after BB checks the flop.”

    Yes, that’s what I understood your table to mean. I myself wan’t clear. What I meant with “how [they] relate (to the donking range)” is that the variations in Snowie’s recommendations may partly be explained by the fact that the BB’s checking range varies among the different board texture since this checking range consists (exactly) in his preflop calling range minus his flop donking range. So we must account for this. If some flops are such that they make the BB’s checking range stronger (or differently balanced), then, the occurrence of those flops in conjunction with the action of the BB’s checking may constitute a “bad event” from our perspective, which justifies us in deviating from average frequencies.

    I’ll comment some more next morning.

    #810

    Sevendeuceo

    See my response on top of yours.

    #811

    Sevendeuceo

    Just another short comment. There are two essential features to a balanced range construction. The correct “defense” frequencies and the correct bluff/value ratios.v (The vocabulary here is fuzzy). As previously noted, there are some branches in the decision tree where we need not be balanced because we are partly giving up. This is compensated by other parts of the decision tree where the opponent has paid a compensating price. Miller stresses the first essential feature but he provides much details about how to go about balancing the bluff/value ratio within the “defense” range. (I use the term “defence” loosely to cover all branches in the decision tree where we aren’t “giving up” — e.g. we are betting, raising, calling, or checking with the intention of check-raising, or checking with the intention of continuing on a later street.

    Now there are four possibilities here, when we see Snowie advocating a very low frequency “defence” range. (1) Snowie is a fish and we can exploit it in that spot. (2) This simply reflects a bad event — so let us analyse why it’s bad. (3) This is not the whole of Snowie’s defence range since Snowie isn’t giving up at all but plans to call or raise later on on that street or thee next ones. (4) Some other defense frequency would actually be better (closer to GTO) than what Snowie recommends but Snowie is constrained to obey the best strategy it has come up with when both himself and his opponent are constrained to only bet or raise 1/2 pot, 1x pot, 2x pot (or shove?). The low recommended frequencies likely are explained by some combinations of those four possibilities.

    I think the main point here, and this indeed may be a small lacuna in Miller’s book (and there ought to be many lacuna in an introductory book) is that while it is paramount that one’s whole defense range be balanced, it is fine to reduce the frequency of our betting or calling ranges below what Miller recommends, and still not become exploitable, provided we aren’t simply giving up on all those extra hands that we check, but rather are thereby creating a stronger and more balanced checking range. In other words, our “defense range” must be close to 70% (equivalently, our “give-up range” musn’t be much smaller than 30%) on pain of becoming exploitable but it is permitted to “slow play” some part of this range and hence spread some part of it (part of the 70%) in other branches of the decision three (not just the call or bet branches).

    #812

    Sevendeuceo

    Let me add just this for now. In the case where Snowie would actually recomment folding significantly more than 33% or the time on any street, then the possibilities mentioned above reduce to two. (1) Snowie is a fish. Or (2) we have reached a node in the decision tree that constitutes a bad event. (This most likely is compensated by value gained in other branches).

    #813

    BNE

    Just thought I’d chime in here.
    I was interested in Snowie after I read Miller’s endorsement of it. At first glance, I was quite excited about it. The first thing that interested me was to chart out Snowie’s preflop ranges. I also played with the trainer a bit and went through their blog. During a couple days of working with it, I found myself very often confused by it’s recommendations and seeming inconsistancies. I finally got around to the 2+2 thread about it, and opinions are not very good over there. I would have thought there would be more people talking about how they worked with it, and how it had helped their game, but no one is really saying that at the moment. There seems to be a lot of doubt in terms of Snowie’s abilities, and the people over there at Poker Snowie do admit it’s a work in progress. I’ve decided to shelve it for now, and watch and see how things progress.
    I did finish my first reading of Janda’s book though, and I’m feeling quite positive about it going together with Poker’s 1%. Right now I’m going through Janda’s 2+2 thread where he has posted about 250 responses to people’s questions. I figure I’ll probably have a lot of the same questions that other people do. The book has some typos/errors in it and he corrects some of them in posts #1 and #376 if you are interested.
    Also wanted to say thanks for posting that back and forth with Ed Miller. It was short but really good to read since the answers are coming right from Ed.
    It’s great the way you authors make yoursleves available like you do.

    #816

    QTip
    Keymaster

    BNE:

    Thanks for your thoughts.

    I’m having a different experience. The more I work with Snowie, the more confidence I have in what it’s suggesting.

    There’s much to filter/dig through with all this stuff. I’m working on it a piece at a time…just need more like 124 hours in a day.

    :)

    #817

    QTip
    Keymaster

    72o:

    Appreciate the thoughts. I think I nodded my head to everything as I read along.

    #887

    QTip
    Keymaster

    Let me add just this for now. In the case where Snowie would actually recomment folding significantly more than 33% or the time on any street, then the possibilities mentioned above reduce to two. (1) Snowie is a fish. Or (2) we have reached a node in the decision tree that constitutes a bad event. (This most likely is compensated by value gained in other branches).

    I was chatting about this with Josh Plotkin. He threw out the idea that perhaps passing up a profitable bluffing opportunity on one street is still best because there’s a more profitable option in the future.

    There are particular situations in snowie where it’s folding 50% on the flop to a 1/2 pot bet. This looked incorrect to me. However, the more I’ve explored these situations, I’m noticing the hands that benefit from donk betting to take advantage of this profitable bluff find more value on a later street and get there often enough to overcome the EV of bluffing the flop. I’m getting more comfortable with what I’m seeing.

    Perhaps a common premise needs challenged. “We must ensure nemesis cannot profitably bluff 2 blank cards on an early street.” This now feels incorrect to me. It makes sense on the river where nemesis has no future options; however, much is yet to happen on early streets.

    If Villain’s most +EV strategy is to not make a +EV bluff on the flop, then we must not devise a strategy to stop him from doing so. In fact, we could send him a hopeful invitation.

    Stay tuned for me to update with an “oh…perhaps I was wrong.” ;)

    • This reply was modified 3 years, 6 months ago by  QTip.
    #906

    Sevendeuceo
    Participant

    “If Villain’s most +EV strategy is to not make a +EV bluff on the flop, then we must not devise a strategy to stop him from doing so. In fact, we could send him a hopeful invitation.”

    I am unsure I understand correctly. This would (perhaps) make sense as an exploitative strategy. But a GTO strategy wouldn’t seek to encourage some specific nemesis strategy but rather (by definition of GTO) seek to minimize villain’s EV for any possible strategy he might employ. A strategy that involves folding some flops with a high frequency might have a higher EV against villains who would take the bait and thereby forego better opportunities on a later street (if that’s what you meant) but it would lose against some counter-strategies that don’t take the bait. But frankly, I’m unsure I can so much as make sense of the idea of folding an early street to prevent a future bluff, so I might be misunderstanding.

    • This reply was modified 3 years, 6 months ago by  Sevendeuceo.
    #909

    QTip
    Keymaster

    “This would (perhaps) make sense as an exploitative strategy. But a GTO strategy wouldn’t seek to encourage some specific nemesis strategy but rather (by definition of GTO) seek to minimize villain’s EV for any possible strategy he might employ. A strategy that involves folding some flops with a high frequency might have a higher EV against villains who would take the bait and thereby forego better opportunities on a later street (if that’s what you meant) but it would lose against some counter-strategies that don’t take the bait.”
    By definition of GTO…let’s make sure we’re on the same page with this definition.

    When I speak of a GTO strategy, I mean Hero is employing one of the strategy pairs in an equilibrium. Hero’s strategy is exploited only when Villain plays the other strategy in the equilibrium; i.e., Villain cannot move his strategy and make Hero’s EV lower than it is when Villain employs the equilibrium strategy.

    At this equilibrium, both players are minimizing the other’s EV. If Hero will not alter his strategy regardless Villain’s actions, then a Hero wanting more money hopes Villain leaves the strategy pair so that Hero’s EV increases.

    When players talk about playing GTO, I believe they typically mean using one of the strategies in the equilibrium regardless Villain’s actions (which is how I use the term). Obviously, when Villain isn’t at an equilibrium strategy and we are, we would improve our EV by moving our strategy, based on the umbrella of game theory.

    To make sure we’re on the same page, when I say someone is playing GTO, I mean to say they’re fixing their strategy to the equilibrium strategy regardless Villain’s decisions; i.e., they’re not attempting to exploit Villain’s strategy.

    “But frankly, I’m unsure I can so much as make sense of the idea of folding an early street to prevent a future bluff, so I might be misunderstanding.”

    That’s more or less what I’m saying.

    This is what I’m thinking at the moment…

    The scenarios in Snowie look something like this.

    Pot $10

    Snowie’s suggestions: BB donk 1% at half pot
    BTN’s response if BB donks half pot: fold 40% of range

    We look at these numbers and think…what? Why wouldn’t BB donk much more frequently than that when BTN is folding so much more often than 33.3%?

    However, the rest of the strategy goes like this:

    BTN cbets 50%
    BB folds 70% to a cbet (say this includes trash hands that could have benefited from donking 1/2 pot).

    However, if BTN checks back the flop:

    BB donks pot with x% on the turn, and this includes the previously mentioned trash hands
    BTN folds 67%

    So, if you look at the EV of those trash hands:

    Flop: For the trash hand, the flop donk has an EV somewhere around 1bb (varies, of course…but just looking at immediate EV based on FE).

    Turn: For the trash hand, the turn donk has has EV somewhere around 3.4bb. The trash hand gets there half the time from the flop, giving it an EV from the flop of @ 1.7bb. This is higher than the EV of a flop donk bet, so if BB is to minimize BTN’s EV, he picks the higher EV decision.

    So, I’ve followed the trash hands through lines in a few scenarios. They are doing at least as good or better as they would in a profitable flop donk. Sometimes they’re used in a flop c/r, sometimes a turn donk or turn c/r. Sometimes a river donk or river c/r.

Viewing 15 posts - 1 through 15 (of 31 total)

You must be logged in to reply to this topic.