Recognizing variance

Take a look at your biggest pots in HOM. If you got sucked out on river with almost all your chips in pot or got coolered preflop/flop, etc... thats variance graph wont show

Yeah those are the kinds of things that i'm talking about. It feels like getting coolered is such a subjective thing. Probably because I lack experience though.

I'm thinking though that maybe i should be thinking about what part of villains range i'm running into when i play a certain line cause that is a part of variance is it not? say if i run into the weaker part of his range and win a big pot then that's an example of good variance?

If it helps I think you will notice this more as you play more. I would try looking into that MBML video on here under the best videos for silver section on pre-flop at six max I found that really useful. I think that the fact that he tells us to stab the btn so much shows how passive low limits are. Also the other video in this section by raskalnikov talking about not getting fancy play syndrome helps.

I wouldn't worry so much about it because if you are worried you played a hand wrong you can always post it in the forum and have people look at it. And yes you are right as the 'good variance' you talk about can come all at once and it will show you just how bad people can play even players that are otherwise quite solid.

Hey Benm

thanks for the input. I'm just reading Tendlar's book at the moment so trying to practise recognizing variance and both mine an my opponents' skill in real time

Variance:

150 BB effective
You hold TT on a flop of 2,4,8r
You bet and are called.
Turn and River are both paint, you both check.

At showdown villain shows 9s.

It's variance when the flop is 2,4,9 with the same holdings.

It's also just variance when you get a walk in the BB while holding AA

So here is how I divided it up in my mind:

It is "just luck" when the outcome is solely due to the deal of the cards.
It is bad play when the outcome is affected by HOW you play.

So:
Slowplaying AA (for example) when you could have bet the flop, turn and river for value each street isn't variance. Whether it is good play or bad depends on the villain, and the situation.

We all get the same cards.
The more skilled player will win more with his winning hands and lose less with his losing hands. The REALLY skilled players can tell one from the other.

Peace,
--VS

You need to think about reciprocity, a word made up by Tommy Angelo. We all get the same cards in the long run. In the situation that Vorpal discusses above, if when you have 99 you can get him to fold TT, but when the situation's reversed, you get to show down your TT, you win. But of course you have to think of it as range v range, not hand v hand. In the long run, poker is like duplicate bridge, but with extra lols.

I suppose one of the first things I recognized, when I actually thought about it again, was that variance was neither good nor bad, it just was.
I think one of the biggest influences is the way I play a hand. If I bet too big, villain folds; if I bet too small, villain calls, that kind of thing. Getting either one wrong can make a huge impact on how hands play out, as does betting/not betting at the wrong/right time.
So knowing ranges helps but I also see an agreement between players which needs to be honoured.

I hope I'm making some sort of sense here, it's a bit rushed.

I'm pretty sure u can't understand variance completely unless u have deep understanding in maths and statistics. Almost every poker player will say that he understands what's variance, but I think it's bullshit. Try to ask some math guy to explain variance in general, it will be the same as u explaining parents that u lost 50k this month, but you're a winning poker player.

How do u think, what's the minimum of hands needed to determine your winrate ? Most players think they're pretty good because he's running like 6bb/100 over 100k hands, but actually it's nothing. For example, if you're 1bb/100 winner, there is ~15% chance that u will be losing or breaking even after million hands. That's the amount many players play a year, and if so, u have a year without profits, but you're still winning player. How many of you will believe this after a year like that ? You can do your own calculations here. Holdem manager give u std dev stat, if u don't have samplesize, it usually will be something like 8.5-9.5bb/hand.

I think you shouldn't worry about variance too much, it's only justification for a bad month. Also think about tournament players, they experience multiple years without profit and it's normal. Then they ship a million or quit ♦

It is "just luck" when the outcome is solely due to the deal of the cards.
It is bad play when the outcome is affected by HOW you play.

This seems too general VP i don't really know how to spot these instances. Because in almost every hand there is the opportunity to make fumbly plays but it seems difficult to see how much variance does play in some spots more than others.

In the situation that Vorpal discusses above, if when you have 99 you can get him to fold TT, but when the situation's reversed, you get to show down your TT, you win.

Yeah i think I understand this but how does it relate to variance?

@ gadget yeah i agree that variance just is and it's something that we all have to get used to but how elusive is it? I think I do need work on my ranges though and it will help me to understand it a bit more.

Yeah I don't really understand how much impact variance does have on the game. I do want to understand it much more though so that I can't mistakenly justify bad plays by just saying "oh its just variance". I want to know about variance so I can be less result-oriented.

Originally posted by legand73

In the situation that Vorpal discusses above, if when you have 99 you can get him to fold TT, but when the situation's reversed, you get to show down your TT, you win.

Yeah i think I understand this but how does it relate to variance?

Variance is related to how long it takes to pick up all those reciprocal edges. I think that once you realize that this is the game that you're playing, you start to appreciate how long it takes for things to even out. However, I agree that it's a bit tangential to the discussion.

However, I agree that it's a bit tangential to the discussion.

I wasn't being a smart ass by the way just genuinely asking. Yeah i've been watching Bogdan's vids on variance and am starting to grasp how much a decent sample size actually needs to be to even it out.

It may never even out for you. ♠_evil:

yeah, scary thought

Originally posted by getdotacom

Holdem manager give u std dev stat, if u don't have samplesize, it usually will be something like 8.5-9.5bb/hand.

could you please explain/interpret this figure in english. lets say in 1m hands you have std dev of 9bb/hand

Originally posted by grohon69
Originally posted by getdotacom

Holdem manager give u std dev stat, if u don't have samplesize, it usually will be something like 8.5-9.5bb/hand.

could you please explain/interpret this figure in english. lets say in 1m hands you have std dev of 9bb/hand

A standard deviation is a measurement of variance. It's symbol is σ, called sigma.

Let's say you are a great player and expect a winrate of 10bb/100. A common standard deviation for a NL 6 max player would be something like 95bb/100.

In a normal distribution (the bell shaped graph) 68% of the time you will fall within 1 standard deviation of the mean (your expected winrate 10bb).
95% of the time you will fall between 2 standard deviations from the mean
99.7% of the time you will fall between 3 standard deviations from the mean.

Imagine each dot in the colored part of the bellshaped graph a 100 hand sample out of billions of hundred hand samples.

i.e. if 1 sd = 95,
over 100 hands you can expect:

- 68% of the time you will will end up with a winrate between 105bb and -85bb
- 95% of the time a winrate between 200bb and -180bb
- 99.7% of the time a winrate betwee 295 and -275bb

Now, what we can do something called a monte carlo simulation which is basically sticking a bunch of these 100 hand samples together randomly. Do it something like 10,000 (100k hands) times. Then do it again and again (keep hitting that calculate button) and again and again and... well you get the point. And you can see different possible journeys.

You can also plug in different winrates and see how dramatically that changes things. You definitely should do it with 5bb/100 and 3bb/100.
There you go. Now you understand NL poker variance.

The tldr version is, when people ask you how much you made playing poker. Don't tell them a sum. Tell them, 'my hourly is x with a standard deviation of 9x'. And then wait. They will either happily say 'nice hourly, good job', and you leave it there. Or if they've done some stats, their eyes will light up and they'll understand.

when people ask you how much you made playing poker. Don't tell them a sum. Tell them, 'my hourly is x with a standard deviation of 9x'. And then wait. They will either happily say 'nice hourly, good job', and you leave it there. Or if they've done some stats, their eyes will light up and they'll understand.

I'm a huge fan of this one

...their eyes will light up and they'll understand.

:f_wink:

Originally posted by jules97
A standard deviation is a measurement of variance. It's symbol is σ, called sigma.

Let's say you are a great player and expect a winrate of 10bb/100. A common standard deviation for a NL 6 max player would be something like 95bb/100.

In a normal distribution (the bell shaped graph) 68% of the time you will fall within 1 standard deviation of the mean (your expected winrate 10bb).
95% of the time you will fall between 2 standard deviations from the mean
99.7% of the time you will fall between 3 standard deviations from the mean.

Imagine each dot in the colored part of the bellshaped graph a 100 hand sample out of billions of hundred hand samples.

i.e. if 1 sd = 95,
over 100 hands you can expect:

- 68% of the time you will will end up with a winrate between 105bb and -85bb
- 95% of the time a winrate between 200bb and -180bb
- 99.7% of the time a winrate betwee 295 and -275bb

Good explanation.
So, if variance is low -- then the curve will be narrow and high, and if variance is high, then the curve will be low and wide.

Is that a correct conclusion?

To put that simply, If variance is low, must 100k runs will have a mean close to the overall mean.
If variance is high, then some 100K runs will be far better than normal and some will be far worse.

So "variance" is not synonymous with "luck", but is better read as "the range of possible results that should be considered normal"

Cheers,
--VS