No announcement yet.

What is Variance? (talk me down)

  • Filter
  • Time
  • Show
Clear All
new posts

  • What is Variance? (talk me down)

    I see talk about variance all the time. Upswings, Downswings, badbeats, rungood. But what is it?

    In the past I thought variance related directly to sort of cards you being dealt preflop. But now I'm realizing I don't know anything about it.

    Is variance just that--the cards you're dealt? Or is it all of it?

    Obviously I'm coming upon this curiosity because I've recently seen a lot of bad beats. I've been playing blitz on ACR to take advantage of the promo. I was doing great! Then all of a sudden, the past 3 days (1k or so) I've just taken beating after beating.

    I'm talking about losing:
    • Fullhouse over fullhouse
    • Fullhouse over nut flush
    • Straight flush over flush
    • Trips Aces out-kicked (J vs Q)
    I'm not gonna lie... I absolutely googled "is ACR rng rigged for action". Oh and I didn't just google it. I dug through archives like a freaking detective.


    Is this variance or is it a lack of skill? Or both? And do you have any good resources/tips for playing through it?

  • #2
    Say you have pocket aces and go all in preflop against pocket kings. Your an 82% favorite in this spot. So for every 100 times this comes up you lose 18 times. Maybe you have a hot streak and win 30 in a row. Or maybe you run cold and lose 8 straight (despite being the favorite). That’s variance. Going all in was the correct decision despite losing 8 in a row. In reality your edge is rarely this big. 60% is pretty good in most situations so you can see how long you could be a loser there.


    • #3
      Poker is a game of incomplete information, part of that information we don't have being the cards that will come out on the flop, turn and river. Variance to a degree includes your pre-flop holdings (as we approach infinite hands we get every combo an equal number of times but obviously if we only play 1,000,000 or 100,000, or 10,000 hands we get some combos more often than other,) but more frequently variance refers to the run outs on future streets. If we get it all in with AA against KK preflop we're ~81% favorite, but variance means we can still lose here 3 times in a row, etc. Variance is just the inconsistency from the predictive probabilities that assume infinite trials to the observed outcomes which are obviously over a finite sample size. Variance could also include when you accurately put an opponent on a range that makes a play profitable, but happen to run into the top of that range and lose more than your equity would indicate you "should" win. That ones tough because you never actually know what your opponents range is.

      Ultimately you just have to develop and execute a strategy that's +EV based on the predictive probabilities, then play allll the volume you can until your observed outcomes get closer to what they "should" be. (Like the strategies at PC.) Winning strategies lose hands, lots of hands, big hands, there is no way to win every pot. Should also be aware not all losses are variance obviously, we make mistakes (although I guess from a certain perspective that's another kind of variance,) we need to be going over hands in detail and spotting the mistakes and correcting them.


      • #4
        Ok so it's more than just what you're holding. That's good news but I guess that means it's even further out of my control. I kinda thought I could just fold variance away but I guess it's not that easy. I see why playing at the proper stakes is so crucial now. Which is another thing I wasn't doing. During the blitz promo I played a little higher that I normally would play online. Felt fine when I was winning. But not so much now.


        • Dilly
          Dilly commented
          Editing a comment
          If you haven't read JL's bankroll bible, definitely check that out.

      • #5
        Variance includes many different factors. But overall it just means that different forms of luck / random chance are going to affect your results in the short term. Over the long run ( eg., 50k hands things tend to even out ) but over the shorter term, e.g, 10k hands there is lots of luck in your results. So looking at any 1k sample and looking at your results is pretty meaningless. Over small sample sizes like that you can review your hands and see if you think you misplayed them. You can look at overall stats like VPIP, PFR, 3bet % and see if your stats are somewhat in line with your goals. But anything related to your win rate is going to be based way more on luck than on skill.

        Just imagine something as simple as if you are dealt KK a lot when your opponent is dealt AA. That's variance. Instead of getting dealt KK when they have QQ, you got it when they had AA. You play the hands perfectly and you lose all of them. Variance is things like being on the river with a A high flush and villain has a flush too so you get paid off.

        Online play is awesome for reviewing your play and plugging leaks in your game. Just don't rely on small sample sizes for win rate.


        • #6

          talk about variance , maybe we can link to Ev, Expected value. variance can be good or bad , as most people only complain about bad no one will talk about them taking more than most their fair share.

          if the probability of you winning that hand is 70/30 , and you only win 5 out of 10 time this is variance this is not good to you. when you winning 10 out of 10 time you are running good and the variance favour you.

          Poker is closely link to Ev, when you hear or seen article talk about this bet or this jam or this call is +ev , +ev $10 , it does not mean this run out hero will win $10, but it just mean in the long run average out per time he will win $10 per time
          but in actual result hero maybe losing 100 or winning 100 etc

          example pot 100 . river villain jam 100 . i have 50% chance of winning i call this 100 jam. . 50% of the time i going win the pot 100 + villain 100 , 50% of the time i going to lose my 100. so my ev = $50

          so in actual fact if i meet this spot 100 time i should be $50 x 100 = i should win 5000,
          but variance dont care i might be winning 5500, i also can be winning 3500, so the bigger your volume and repetition , your result should reflect the Ev

          then what is a good volume sample , we are looking at 100 000 to 1 mil hand sample size. but if you play live game this type of volume is hard to achieve .


          • #7
            Right, I think I get it now. I guess I just never thought much about it because I always assumed it related directly to the cards I'm dealt and nothing else. Now I understand there's more to it.

            Over this past week I put in about 10k hands and I will definitely say that I was running good. All my AA and KK held, and overall I knew I was running better than I should be. But that last 1k really made me realize that I didn't quiet have a grasp on this beast we call variance.

            I assumed hole cards were my variance and the rest was up to skill. But I see now that's not always true. I'm not sure why I thought that way but I'm glad I learned more about it sooner rather than later.


            • #8
              Variance comes from mathematical statistics. You can model it, but that is unnecessarily complicated.

              It's basically how much you are expected to deviate from the mean (expected value).

              A lot of people do not understand variance and then blame Pokerstars or other poker rooms for having bad beats. Just don't be that guy


              • Cloutier
                Cloutier commented
                Editing a comment
                So we have EV and THEN we have variance. So if I’m 80% in a spot, variance could mean that there’s a deviation away from even that number.

            • #9
              If you use software to track your hands then you can find out how well you've been playing without variance to some extent.

              Over a sample I was looking at recently I was running at 20bb/100. Which means for every 100 hands I was playing I was winning 20bb - this is a huge win-rate.

              I then took a look at my all-in equity adjusted win-rate which was a lot lower at 5bb/100.

              What this figure shows is how I would've run if there was no variance in poker and if I got it in as an 80% favourite, then I would win exactly 80% of that pot. Basically I had been running better than what my equity would suggest, i.e. running with positive variance.

              There are a lot of variables that this stat does not take into account and as such it's never going to be totally accurate but it's still interesting.

              My hand sample was just a few thousand too which also detracts from the accuracy which is something else to keep in mind.


              • Cloutier
                Cloutier commented
                Editing a comment
                Yeah I have pt4 and I’ve seen a couple videos talking about this feature but never dig into it. I will definitely be doing that tonight. I think I’ve got 50k hands total so at least it’ll give me something to look at.

                I should’ve researched all this stuff sooner. It’s very interesting... and a little aggravating 🥴

              • LondonImp
                LondonImp commented
                Editing a comment
                50k is a fairly decent sample size. Not perfect, but you should be able to get a fair idea of where things are going.