Why Randomizing Kills Your Winrate

Поділитися
Вставка
  • Опубліковано 4 гру 2024

КОМЕНТАРІ • 70

  • @maxwelllittle5291
    @maxwelllittle5291 Місяць тому +61

    Trying to be unexploitable is a waste of time when very few players would even notice your frequency error, and even fewer would be able to counter exploit.

    • @joebarra5273
      @joebarra5273 29 днів тому +1

      It depends on the game. In deep stack cash it is important to have nut hands at every node. If you always bet (or always check) flush draws (for example) even losing 2/5 players will eventually figure that out.

    • @torsti4791
      @torsti4791 27 днів тому +1

      I agree, unless you're at a high level you should play a solid style that works for you and exploit others as well as adjust yourself based on reads

    • @AAZFortnite
      @AAZFortnite 9 днів тому

      @@joebarra5273no they won’t lol

  • @GTOWizard
    @GTOWizard Місяць тому +8

    Fascinating video!

  • @BoXianOng
    @BoXianOng 29 днів тому +1

    Great video! Thanks for helping us see the forest instead of just the trees and yes - most opponents in our games cant even spell balance, let alone figure out how to exploit any imbalance on our part. Thank you for the video!

  • @chrislee-r2i
    @chrislee-r2i 5 днів тому

    Love your videos, you make it easy to understand.

  • @samuraijack1371
    @samuraijack1371 Місяць тому +4

    I think what's even more interesting is that JxTc should be actually bluffing on the river at a decent frequency and I think that is a better exploit than X/F
    -you have poor equity to begin with. Under GTO is around 37% (low end of your SDV range). In practice you could argue that IP's range is loaded with Ax,Kx, QQ/JJ. IP isn't betting thin enough on turns on a wet board. Though they also don't slow play sets and 2Ps but overall I would say JT doesn't even have 37% equity on river.
    -Having said IP doesn't protect turn x-back with 2P/sets/starights, betting 150% with JTc is excellent a will over perform GTO. its blocking flushes, sets and straights. making the nuts in IP's range even rarer.
    -X/R JTc could be an alternative but it think its not very good in practice, Population doesn't bet thing enough and when they do they often tend to station citing some blocker reason. On this board IP has tons of Pair +nut blocker hand in the river B range that will not fold to a xraise but should under GTO.
    so of the 3 options, OB bluffing is the best compared to x/r or folding.

    • @gavilanu
      @gavilanu 29 днів тому

      totally agree. IP player is supposed to check half of his 2P in this turn , KT is a pure check and almost nobody plays like that, at least on mid-stakes

    • @indjke
      @indjke 29 днів тому

      I’m low stakes scrub but he said correctly that check back turn shows weakness and then does not attack him when flush closes…
      Could we potentially force opp to fold some stronger hands like QT or KJ if we bet river?
      Or get value from 88 99?

    • @TeslasMoustache419
      @TeslasMoustache419 29 днів тому

      You could try an overbet and get him to fold a pair. It's not the end of the world if you get called. It's even good advertisement if you can get the opponent to be suspicious and sticky later in the session (in a larger pot when you have the goods). I actually like to attack smaller pots relentlessly when fish show weakness for this reason, attack until you get caught and then change gears.

    • @PokerGiraffe
      @PokerGiraffe  29 днів тому +2

      I think we might have a bit more SDV in practice, because population might not bluff with as many weak pairs when checked to. This does increase the EV of checking, along with the fact that HJ might arrive on the river with more weak pairs than they should (instead of barreling turn sometimes).
      But what you say makes sense as well, so at the end of the day it might just be a close decision.
      I actually think check jamming is okay given how fast HJ checked back turn. Surely Axcc would have thought for a while longer!

  • @cc-yf5ck
    @cc-yf5ck Місяць тому

    i wish you wouldn’t release this so late my time, now i’ve to wait a day to watch this! can’t wait, Poker Giraffe!!

  • @Doblou13
    @Doblou13 20 днів тому

    Great video! Did you get Uri's bit from one of his courses? Where can I find it?

    • @PokerGiraffe
      @PokerGiraffe  15 днів тому

      It's on his youtube channel! Search for "understanding poker solvers"

  • @sixwaveholddown
    @sixwaveholddown 26 днів тому +3

    I’m just learning how to play poker and it’s infinitely more difficult than I thought. Like what even is this video

    • @PokerGiraffe
      @PokerGiraffe  26 днів тому

      😂

    • @giovanni679
      @giovanni679 9 днів тому

      you're probably studying a few levels deeper than you should then haha

    • @sixwaveholddown
      @sixwaveholddown 9 днів тому +1

      @ I realized that hahaha. I get it now though I just had to sit down at a table and actually play. So much to learn though still

    • @TooMuchInternetTMI
      @TooMuchInternetTMI 8 днів тому +1

      ​@sixwaveholddown 5 minutes to learn the rules, a lifetime to still not master it. This is kind of advanced poker theory it won't make sense to new players

  • @dr_manuel_graeber
    @dr_manuel_graeber Місяць тому +3

    Is the EV always the same for both possibilities if a solver suggests randomizing between 2 actions?

    • @archie2281
      @archie2281 Місяць тому +11

      Within 0.1 ev at a nash equilibrium but in a real game the decision could be a 30 bb decision with exploitative facts

    • @PokerGiraffe
      @PokerGiraffe  Місяць тому

      well said :)

    • @dr_manuel_graeber
      @dr_manuel_graeber 29 днів тому

      @@archie2281 thank you!

    • @samuraijack1371
      @samuraijack1371 29 днів тому

      @@archie2281true. But it depends on your solver settings. You can set your turn and river accuracies to 0.00001. They solve pretty quick and it’s the river nodes where you need the highest accuracy cause frequencies shift dramatically.
      Though none of this matters at low/mid stakes and even at say 5K NL

  • @zinhaopoker
    @zinhaopoker 27 днів тому +1

    you're the best! 💙

  • @brandonnel4687
    @brandonnel4687 21 день тому

    Fedor and bryn was beaten by 2 recreational players known as ossi and vlad at the final four in the triton $150 k buy in even some notable names such greenwood,chidwick and Dan smit was at th FT.

  • @leonardoribas3997
    @leonardoribas3997 28 днів тому

    Great video! A question: does that apply for MTT as well, or there're other nuances? Cause in MTTs this seems even less important for the fact there're so many variables such as RP, drop equity, stack distribution

    • @PokerGiraffe
      @PokerGiraffe  28 днів тому +1

      Not an expert on MTTs, but I don't see any reason why the logic wouldn't apply 🙂
      Live MTTs especially are notoriously soft, definitely wouldn't make sense to be randomizing.

    • @leonardoribas3997
      @leonardoribas3997 28 днів тому

      @@PokerGiraffe Sorry, I forgot to mention that I was asking about the concept of not needing randomize on later streets because of lack of vision from our strategy by population. Like, on MTTs this may apply even for flop and preflop maybe?

    • @PokerGiraffe
      @PokerGiraffe  26 днів тому

      It really depends on what you think you can get away with. In a lot of softer games you don't even have to randomise preflop, because players might not even recognise your frequency mistake, or might not be exploiting it in the right way.

  • @MDA_Demon
    @MDA_Demon 26 днів тому

    We can use MDA to help us make a maxEV decision with our hand that doesn't rely on RNG

  • @RunItTw1ce4858
    @RunItTw1ce4858 28 днів тому

    9:00 no made hand should be called bluff allowance

  • @hongseongi3949
    @hongseongi3949 Місяць тому

    What a nice video! Thanks

  • @BigG-d6t
    @BigG-d6t 24 дні тому

    In futuro potresti approfondire anche gli exploit preflop,grazie

  • @theworldisdamned7991
    @theworldisdamned7991 Місяць тому +1

    Shotout to uri peleg❤

  • @consumer61
    @consumer61 29 днів тому +2

    Good video. I'm not sure I agree with your example, however. You said a maniac who bluffs 100% (i e. overbluffs) would win money from someone playing close to gto but overfolding by 0.1%...
    I don't think that's accurate. He will make money in the lines where we overfold, but he is burning money in the lines where he oberbluffs and we correctly call down. The latter is way more frequent than thr former, hence we will profit massively versus this guy over any decent sample (assuming fixed strategies).

    • @PokerGiraffe
      @PokerGiraffe  29 днів тому

      Overfolding by 0.1% meaning folding 50.1% instead of 50% haha. So we still fold more often than we call.

    • @consumer61
      @consumer61 29 днів тому

      @@PokerGiraffe Right, but the point remains

    • @PokerGiraffe
      @PokerGiraffe  29 днів тому +2

      His bluff is +EV by definition, because we are overfolding. Yes he will get called down sometimes, but not often enough to make his bluffs -EV. If his bluffs were -EV, then by definition we would be overcalling not overfolding.

    • @consumer61
      @consumer61 28 днів тому +1

      @@PokerGiraffe I disagree. His bluffs are not -ev because we are overcalling, they are -ev because he is overbluffing...
      Yes, he will make more money than he should in the lines where we overfold, but most of the time we are defending correctly, meaning we are making way more money than we should by running into bluffs and winning much more frequently than gto.
      This far outweighs the times his overbluffing makes money from our 0.1% overfolding. In other words, his deviation from gto is far greater than ours.

    • @PokerGiraffe
      @PokerGiraffe  28 днів тому +1

      Let's say villain bets $100 into a pot of $100:
      If we call 50%, EV of villain's bluffs are
      0.5(100) + 0.5(-100) = 0
      If we call more than 50%, villain's bluffs are -EV
      If we call less than 50%, villain's bluffs are +EV
      EV of bluffs has nothing to do with bluffing frequency itself - instead, they depend solely on the defender's calling frequency. This is why if you nodelock the defender to overfold slightly, solver adjusts by bluffing 100% of air (see 9:24). Which obviously wouldn't be the case if bluffing was -EV.

  • @adrianoalves20
    @adrianoalves20 29 днів тому

    Awesome. How about a play and explain video exploiting everyone on low stakes?

  • @にのまえつち
    @にのまえつち Місяць тому

    Nice video thanks!
    What is the random number application in the video0:21?

  • @男特
    @男特 27 днів тому

    good job man update more❤

  • @oneone_isme
    @oneone_isme 29 днів тому

    This video is significant cause it explained why it's difficult and unnecessary to emulate GTO,if you get one handclass wrong,and it messes everything up;But the main question is why did this 11-minute video take you three months?

  • @VinnyPoker
    @VinnyPoker Місяць тому

    Amazing video aswell

  • @kirillkaramelniy7360
    @kirillkaramelniy7360 Місяць тому

    U need 2 know frequinces to understand h much value/bluffs should be in the spot and exploit it by deviating from gto if u see ur opp put too much value/bluffs on the street
    And if u know that ur opp is maniac or smth and bluffs 100% air,that means u never fold and this is pure gto

  • @illegitimate0
    @illegitimate0 Місяць тому

    You have to be balanced when playing a computer that's seen you play thousands of hands. Oh, but you're not doing that? You're playing drunks at your local casino who want to see a flop? Then don't.

  • @yoniker83
    @yoniker83 27 днів тому +2

    Giraffe, You claim that using RNG can hurt the win rate at lower stakes, and give an example of a "close to GTO" player losing vs a completely imbalanced player. While the second part is provable, I wonder about the RNG part, especially on the flop and turn.
    Also, you lay out a simple "3-bet or fold" preflop strategy for low stakes, which is completly imbalanced and the only exploit is for villians to start 4-betting light.
    All of this makes me think, how much is like "in depth" knowledge of GTO is relevant for lower stakes? What's your opinion on this? Like, am i wasting time by studying GTO in depth? (eg learn the bet sizing and global frequencies, local frequencies for different hand classes on different textures, how GTO responds to check-raise etc)? :(

    • @PokerGiraffe
      @PokerGiraffe  26 днів тому +2

      Studying GTO is important, but you always have to bear in mind that you're doing it in order to exploit better.
      Just like in the EP vs BB example, where we saw that solver is turning lots of pocket pairs into bluffs - the point is not to bluff with these pairs ourselves, but rather to recognise that many players are not finding these bluffs, and think about how to exploit them.

  • @kaaristotelancien3005
    @kaaristotelancien3005 29 днів тому

    In a sense, GTO learn you to exploit better

  • @BigG-d6t
    @BigG-d6t Місяць тому

    Complimenti

  • @brandonnel4687
    @brandonnel4687 29 днів тому

    Nice video, but if you tell us not to play GTO then show us how to exploit the recreational.

  • @NikolijaCryptoPR
    @NikolijaCryptoPR 14 днів тому

    Hey! Great video, thanks for all the help! Is there a way to contact you directly? Please let me know :D

    • @PokerGiraffe
      @PokerGiraffe  12 днів тому

      hey, feel free to leave a message on my website.

  • @uniquelolapps7318
    @uniquelolapps7318 29 днів тому +4

    You still didnt give a good argument why randomizing is killing your ev. You and Uri Peleg have only said that if villain is maniac with 100%vpip you shouldnt randomize, if someone cant adjust vs such a player should just stop playing poker