Hull. Questioning The 1%,

Doug Hull revisits they key ideas from the frequency-based approach to NLHE described in "Poker's 1%." Using solvers that were not available when this book was first published, Doug explains where modern theory deviates from the original recommendations of Poker's 1%, and reviews where the advice from that book still applies.

Please leave questions and comments on this video below.
Moderation In Moderation

Comments

  • eselspieleselspiel Red Chipper Posts: 41 ✭✭
    edited December 2018
    FY!,,,,for the first example Doug verbally says he is looking at double barreling (10:18), but the tree has the flop going check-check.

    Interestingly and maybe coincidental, Adam Wheeler did a video a few months ago looking at ~ 180 different flops and found that a 1/3-pot flop bet was made ~ 70% of the time.
  • TheGameKatTheGameKat Posts: 1,758 -
    Doug's video uses the pot-pot-pot tree of Ed's book. 70% for 1/3 pot is likely consistent with Doug's results.
    Moderation In Moderation
  • ericmwhiteericmwhite Red Chipper Posts: 34 ✭✭
    I’ve spent the last month going through The 1% video series, using it as the basis of my study and changing my strategy to that frequency based model, so I’m really curious if or how how that course might change as a result of these findings?

    @TheGameKat It always struck me that the continuation percentage is elastic based on bet size. Is that the point you were trying to make, that at a lower bet size would increase the percentages in Doug’s results?
  • TheGameKatTheGameKat Posts: 1,758 -
    ericmwhite wrote: »
    I’ve spent the last month going through The 1% video series, using it as the basis of my study and changing my strategy to that frequency based model, so I’m really curious if or how how that course might change as a result of these findings?

    @TheGameKat It always struck me that the continuation percentage is elastic based on bet size. Is that the point you were trying to make, that at a lower bet size would increase the percentages in Doug’s results?

    Yessir, exactly so.
    Moderation In Moderation
  • Doug HullDoug Hull RCP Coach Posts: 1,684 -
    edited December 2018
    The c-betting percentage is, not surprisingly, a function of the following inputs:
    • Your range
    • Opponent's range
    • Flop
    • Stack to Pot Ratio
    • Intended bet sizing progression

    Roughly this is a five dimensional problem, only one variable is mappable to a single number scale (SPR). The intended bet sizing progression is reasonably small set of expected bet sizes. The flop is a large set of possibilities that can be lumped into a set of smaller representative flops. The ranges are huge (13x13)^2 sets that can roughly be modeled as [0%-100%] ranges with some argument as to the ordering of the 13x13 components.

    All this is to say that Miller's book was an attempt to slice through this huge five dimensional hypercube to give us a general look at this single number: c-betting percentage.

    It would be surprising if a single number could be a decent approximation for this complicated of a dataset. What we see with GTO+ on the turn is that even when we hold all five of the above variables fixed and just look at different turn cards we get a huge variation in double barrel percentage.

    So, 70% is not the right one size fits all percentage. Given modern solvers, I am guessing it is more like 50%, but the truth is there is no one-size fits all c-betting percentage. It depends.

    We should not look at this and say "oh, 70% is wrong all my study is useless in this area." We should be saying: "Frequencies are important, that is a huge and important revelation. Let's see if there are patterns that tell me when it should be high, and when it should be low..." Ross is working on a project along these lines and it is really shaping up.

    Co-founder Red Chip Poker,
    Author Poker Plays You Can Use
    Author Poker Workbook for Math Geeks
  • PUPPYPUPPY Red Chipper Posts: 10 ✭✭
    edited December 2018
    [Edited for grammar so I can parse this- Doug]
    Its been beneficial. I've never worked s closely with my ranges and flopzilla as I have when considering a frequency based approach.

    I have said a lot of credit needs to be given to events and the frequencies you choose. I like what was said about the solver and in a 11/90 split the low % action could be used in situations, where it was an option, based on the opponent or other factors of the situation.

    Should players be looking more towards meeting their MDF and adjusting that for hand analysis/frequency awareness?

    Should we be most concerned with solvers now and figuring out these spots and why the solver suggests certain actions?
  • ericmwhiteericmwhite Red Chipper Posts: 34 ✭✭
    Doug Hull wrote: »
    We should not look at this and say "oh, 70% is wrong all my study is useless in this area." We should be saying: "Frequencies are important, that is a huge and important revelation. Let's see if there are patterns that tell me when it should be high, and when it should be low..." Ross is working on a project along these lines and it is really shaping up.

    I didn't finish watching your video and worry that I had just wasted a bunch of time. I am far more motivated by being a part of the 1% number than blindly following the 70% number as some sort of golden ratio. It is clear how a frequency based approach to poker is the best way to win.

    I do remain curious and excited about how your findings could influence the work I'm currently doing. It sounds your opinion is:
    1. When building continuing frequencies, start with 50% rather than 70% (a number that is still variable based on good/bad events, bet sizing, ranges, board texture, etc)
    2. Other than that all the principles and tactics in both Ed's book and the @SplitSuit course stand the test of time (so far)
    3. Ross is working on a project that I should be excited about and offer my time and attention as a beta tester or early reader if you'd like student-level input :-)

    Thanks for the video and the quick response on a holiday weekend!
  • Leandro_1Leandro_1 Red Chipper Posts: 4 ✭✭
    So would have sense to aim to have a 70% continuation range. So the 70% can be a combination of c-bet - check call (brick) turn and check raise (brick) turn IP, and donk - check call - check raise OP in order to work that ranges off-table ?
  • driller1driller1 Red Chipper Posts: 25 ✭✭
    I bought James Sweeny's 1% course. I also have Miller's book. I'm pretty old so I don't know if I have enough time left to do all the analysis, much less do it so much that it sinks into my brain enough that I make the correct moves (if they still are) instinctively.

    It looks like from your analysis that especially for pot sized bets, 70% is just wrong. Is that fair?

    Do you vary the size of your C-Bets much? Also I see one of the variables is opponents range. What about assessing the type of player?
  • driller1driller1 Red Chipper Posts: 25 ✭✭
    Now that GTO+ has blown up Ed's 70% frequency, can you comment on Poker Snowie?
  • Doug HullDoug Hull RCP Coach Posts: 1,684 -
    @Leandro_1 Huh? I do not understand what you wrote.

    @driller1 In a word, 70% is "wrong." Any single number applied to all situations would also be wrong, so this says nothing bad about the original 70% number.

    The 70% was estimated based on a variety of ideas that helped to give upper and lower limits on the percentage. Without the GTO tools that have been developed since publication of 1% this was a very reasonable estimate and pushed the community in the right direction.

    In speaking with Ross, who should be credited with heavily influencing my thoughts on all of this, we went through this process.

    If I had to pick exactly one number to use all the time, like Miller did. I would choose 50%. This is what an engineer would call a "first order approximation." Not super accurate, but it starts to get us to an answer and we can refine.

    Of all the things I listed in my earlier post as factors that influence this c-betting percentage, the variable that changes most from hand to hand and has the biggest influence on the number is the flop. On any given flop, I would want the ability to pick a c-betting percentage from the list of 0%, 50%, 100%. This is what we would call a "second order approximation" The way we think with this is more nuanced, but now we have three possible answers instead of one. We are looking at one more variable though.

    Ross' work has helped to identify the poles, the 0% and 100% boards. The rest are 50% boards. Our approximation has gotten even better with this heuristic.

    Bet size progression, just like c-betting percentage are mostly a function of the flop texture. 1/3pot through pot sized can be reasonable depending on the texture.

    As far as poker Snowie goes, it has many of the same results at GTO+. GTO+ is meant for this work, but you can not play against GTO+. Snowie is meant to be played against. The analysis helps to understand why Snowie agrees or disagrees with your move. The information on the GTO aspects are there in Snowie, but harder to look at systematically.

    I would say these two programs are based on similar analysis but have a different front end and user experience. I think there is a place for both of them in your library. I know I got them both.

    -Doug

    Co-founder Red Chip Poker,
    Author Poker Plays You Can Use
    Author Poker Workbook for Math Geeks
  • driller1driller1 Red Chipper Posts: 25 ✭✭
    Thanks for your reply, Doug

Leave a Comment

BoldItalicStrikethroughOrdered listUnordered list
Emoji
Image
Align leftAlign centerAlign rightToggle HTML viewToggle full pageToggle lights
Drop image/file