Variance

T

The_Pup

Rock Star
Silver Level
Joined
Oct 27, 2009
Total posts
254
Chips
0
Variance is a word that is bandied around poker circles, especially cash game players, with a sense of 'things go up and down but average out in the long run'. This is a fairly good working definition but only scratches the surface, so I thought I might expand on the topic. No doubt someone has done the same here before (and probably made a better job of it) so forgive this relative newcomer to CC if I am raking over old ground.

Let's suppose we are a decent poker player at our chosen level. A session of say 100 hands will have its own ups and downs but our ability is such that we can regard each session as being like a coin flip in which we win $12 on heads and lose $10 on tails. Now that is a profitable game, our expected profit is $1 every time we sit down to play - we are $1 per 100 hands more skilled than our opponents. Now we have a simpler game of coin flipping where we get $12 on heads and pay $10 on tails.

We can take this new game and play it 100 times - this represents 10,000 hands of poker. Our expected return on these 100 flips is $100 since we expect to win $12 50 times and lose $10 50 times. Whilst 50 heads and 50 tails is the average result of tossing a coin 100 times we would not be surprised at 49/51 or 48/52 - this is where variance comes in. In order for our game to remain profitable we need to score more than 45 heads, less than that and our losses outweigh our wins.

It turns out that about five sixths of the time we can expect to throw 45 or more heads - the more heads the more profit, throw 60 and we are on a great run, 70 and we are poker gods. But of course throw less than 45 and we are losing money and that will happen one sixth of the time.

What this means is that you might have a winning strategy and be $1 per 100 better than your opponents but don't be too surprised when you lose money over a period of 10,000 hands. And if you do it is quite possible you won't get your losses back on the next 10,000 or the next. Of course once in a while some poor sod throws 95 tails and goes on a horrible bad streak whilst someone else throws 95 heads, and calls himself 'durrrr'. ;)
 
U

Ubercroz

Visionary
Silver Level
Joined
Jun 8, 2009
Total posts
653
Chips
0
Great thread. Makes me feel better, now I can just keep blaming variance for my losses. "Yeah I've lost money over my last 100,000 hands, but thats probably within a couple standard deviations of normal anyway, right?" Just kidding, yeah good post and something I think people should be more aware of.
 
thepokerkid123

thepokerkid123

Visionary
Silver Level
Joined
Jul 31, 2009
Total posts
917
Chips
0
The problem with variance is that everyone thinks they're the guy who threw 95 tails when they're losing and think they're throwing 45 heads when they're throwing 55.
 
ckingriches

ckingriches

Lucky Multiple League MVP
Silver Level
Joined
Jul 27, 2009
Total posts
2,315
Awards
9
Chips
1
I'm not so sure that poker outcomes follow the normal distribution upon which your theory appears to be based. However, I agree that variance is being used by poker players to "explain" that their short-term ups and, more often, their downs, are more a reflection of probability distribution effects than skill level.
 
T

The_Pup

Rock Star
Silver Level
Joined
Oct 27, 2009
Total posts
254
Chips
0
Thanks for the positive comments and for spotting we are talking about the normal distribution and standard deviation - I was unsure how technical to get with the maths.

I'm not so sure that poker outcomes follow the normal distribution upon which your theory appears to be based. However, I agree that variance is being used by poker players to "explain" that their short-term ups and, more often, their downs, are more a reflection of probability distribution effects than skill level.

You are right that poker (hand) outcomes won't follow the normal distribution but I have reduced the poker play to coin flips with a probability of 1/2 which does follow the normal distribution. The payout is skewed 6 to 5 in our favour because we are a skilled player so the complex probabilities of poker play are already taken care of.

I think that the thing these numbers show is that as a skilled player who is one dollar better than the tables we sit at we should not be in the least bit surprised if we find ourselves on an extended losing run - indeed we should be surprised if it doesn't happen. All it takes is for five 100 hand sessions to go bad.
 
RoyalFish

RoyalFish

Rock Star
Silver Level
Joined
Jan 14, 2009
Total posts
240
Chips
0
I'm not so sure that poker outcomes follow the normal distribution upon which your theory appears to be based.

Individual trials are not normally distributed (I think they'd be Bernoulli), but any sufficiently large number of repeated trials of any distribution are normally distributed. Central Limit Theorem.

RF
 
No Brainer

No Brainer

Losing keeps me sane
Silver Level
Joined
Aug 27, 2009
Total posts
1,853
Chips
0
how many losing coinflips do we add in to take care of the rake?
 
slycbnew

slycbnew

Cardschat Elite
Silver Level
Joined
Aug 8, 2008
Total posts
2,876
Chips
0
Great thread. Makes me feel better, now I can just keep blaming variance for my losses.

pokerkid refers to this point, but keep in mind this also means that we can attribute our winnings as well as our losses to variance as well...
 
T

The_Pup

Rock Star
Silver Level
Joined
Oct 27, 2009
Total posts
254
Chips
0
pokerkid refers to this point, but keep in mind this also means that we can attribute our winnings as well as our losses to variance as well...

Excellent point, Sly. A fact that has uncomfortable implications for some:

Using my model of coin flipping with a skill advantage of $1 we are in profit if we get 45 or more heads. Equally, there are some players out there who have a skill disadvantage of $1. Because of their lower skill level they need to get a much higher number of heads over the 100 flips. It is nothing that remarkable if they get 60 heads and turn in a profit. We can imagine a poor player who starts playing online and soon racks up 10,000 hands and returns a profit - because of variance. There will be a feedback circuit that will tell them they are a good player so they crack on playing the same style of game. The next 10,000 hands sees them getting 50 heads and so they lose. These loses they blame on variance, having a bad run of luck. Since most people believe they are better than most other people at any given task it is easy to slip into the trap of not analysing one's own game but put losses down to variance - 'after all, I did make a profit in the first 10,000 hands.'

I can talk about this with great authority as this was what happened to me when I started online!
 
RoyalFish

RoyalFish

Rock Star
Silver Level
Joined
Jan 14, 2009
Total posts
240
Chips
0
Most people think they're better than average, when obviously "most" people can't possibly be. It's more rule than exception that winning players think they're good and losing players think they are unlucky.

RF
 
T

The_Pup

Rock Star
Silver Level
Joined
Oct 27, 2009
Total posts
254
Chips
0
Most people think they're better than average, when obviously "most" people can't possibly be. It's more rule than exception that winning players think they're good and losing players think they are unlucky.

Depending on your definition of average, it is easy to have most people better than it.

I think your second point is true; though whether you are a 'winning' or 'losing' player is hard to tell from looking at results over 10k or 20k hands. The potential danger for a poor player is to believe their profit is down to skill where they have just got the good end of variance - their bad play is ingrained as 'good play' because they won with it once.

As for me, I'm good enough to know I'm not as good as I think I am.
 
RoyalFish

RoyalFish

Rock Star
Silver Level
Joined
Jan 14, 2009
Total posts
240
Chips
0
Depending on your definition of average, it is easy to have most people better than it.

This is true, but it'd be a shame to derail a good thread arguing that point. Yes, average has multiple definitions. Most people think they are in the top 50% when asked to rank themselves. Better?

RF
 
T

The_Pup

Rock Star
Silver Level
Joined
Oct 27, 2009
Total posts
254
Chips
0
This is true, but it'd be a shame to derail a good thread arguing that point. Yes, average has multiple definitions. Most people think they are in the top 50% when asked to rank themselves. Better?

Yes, I agree. I was worried about being too pedantic when writing the previous post.
 
T

thepit89

Enthusiast
Silver Level
Joined
Jul 15, 2009
Total posts
39
Chips
0
Most people think they're better than average, when obviously "most" people can't possibly be. It's more rule than exception that winning players think they're good and losing players think they are unlucky.

RF
Completly correct, almost any poker player I've ever met said they are better than the average and none of them realize that they might / most likely are wrong.
 
T

The_Pup

Rock Star
Silver Level
Joined
Oct 27, 2009
Total posts
254
Chips
0
how many losing coinflips do we add in to take care of the rake?

Good question. The rake is taken care of in the assumption that we either win $12 or lose $10 over 100 hands. The probability that we win $12 is 1/2 and that we lose $10 is 1/2 - including the rake. The meta-game of coin flipping can thus be treated as a very simple game that makes the maths easier to deal.

Out of interest I did a trial of 100 coin flips several times over (using a virtual flipper - I have got a life). Though the expectation is to get 50 heads and 50 tails, in reality I got 44/56, 48/52, 60/40, 53/47 and 42/58. By my reckoning, this series of flips has two losing runs, one just better than breaking even and two highly profitable.
 
Top