My question has probably been answered before but through searching hte forums I am unable to find the answer I am looking for. I was wondering what you guys think about the Variance rate at 50bb tables vs 100bb+ tables. Are we going to experience a higher variance rate at the 50bb tables or at the 100bb+ tables? Or is it neutral and the amount of bets in your stack don't make a difference in terms of variance?

I found that in 50bb poker, there is less poker played per se but there is more raise-3bet-shove.

100bb should have more variance. Larger pots infer a larger deviation from the mean

you have it backwords, less bbs result in less play and more pf variance. Think of it this way, why is there more variance in hyper turbos than regular sngs?

This is what I thought. I had a guy tell me that the "law of large numbers" states that the more blinds in the game the higher the rate of variance which I totally disagreed with. You would think that playing deeper poker there would be more poker to be played instead of just a push/shove fest which is 20-50bb poker.

Less bb's in play doesn't mean more variance, means less of an edge. So it's just perceived variance, but the deeper the stacks the larger swings you will have from your expected winrate.

yes but variance is not only how large the swings are but also how often they occur.

Don't see what the difference is between 50bb and 100bb in terms of frequency of swings. Should be about the same. If you have a hand worth playing for 50bb in a 50bb game, it's usually the same for 100bb game.

I once summarized it like this to a friend:

In higher BB games, you are likely to experience:
- a bigger influence of skill, as decisions are for more bb and +EV decisions will be worth more. Also someone used to a 50bb game simply has not learned how to play a hand to the river without decisions becoming trivial.
- a bigger influence of variance, as the normal divergence from your EV is just bigger. The difference between (+70bb vs -30bb) and (+210bb vs -90bb) is easy to see.

On the other hand, in 50bb games you will see a lot more all ins before the river, so it's actually a balancing between difference in bb's and frequency. I imagine you'd need significant samples to demonstrate the optimal buy-in amount for a particular style.

Good points in this thread. I always thought that the smaller the effective stacks the higher the variance, good points both ways.

I guess technically more $$ is at risk with bigger stacks, but is there not more overall variance in the shorter games? A smaller edge with smaller stacks just naturally seems like it would have a lot more variance than a bigger edge with bigger stacks.

Doesn't bigger stacks mean bigger edges which means less variance if you are a winning player?

A bigger edge does not mean less variance, typically, no. It means less RELATIVE variance, in the sense that the ratio between your standard deviation and your win rate is smaller, which is good and leads to a smoother looking graph, which is what a lot of people (wrongly) mean by lower variance.

To me its pretty obvious:

Consider just flips for stacks, at 50bb tables you are +/- 50bb/hand, at 100bb tables +/- 100bb/hand, this is pretty clear. Assuming that the average pot is bigger at 100bb than 50bb, the variance will also be higher. Intuitively I'd say the 100bb table would have bigger avg. pots but thats just a guess, maybe someone can confirm?

EDIT: Guess it also depends on the volume of allins too, still think 100bb games would be higher variance

Why is everyone trying to 'common logic' the answer out instead of getting a sample of standard deviations from players that have played a statistically significant number of hands at 50bb and 100bb tables?

Without pulling up my db, I'm fairly sure my standard deviation at 50bb tables has been in the 42 to 45 range over the last 200k hands. Anyone else?

Ave at 100bb is like 80bb/100

Make sure you're using the same units, ptBB or big blinds

yes, and everyone is ignoring this fact.

same reason variance is lower at deep NBI husngs vs turbos.

No, they aren't ignoring this, because it isn't true. A bigger edge does not mean less variance.

A biggeer does not mean less variance.

Learn the normal distribution and calculus guys

I'm a long time reader but I know something about math so I thought I could chime in here.

The variance of a random variable X can be calculated easily as E[X^2]-E[X]^2, where E[something] is the expected value of something.

If you model the games as flips with 60% equity, the game with bigger pots has bigger variance.

In a game with pots of ~50bbs we have a variance of [(.4)*2500+(.6)*2500]-[(.4)*50+(.6)*50]^2 = 2490.

In a game with pots of ~100bbs we have a variance of [(.4)*10000+(.6)*10000]-[(.4)*100+(.6)*100]^2 = 9800.

The resulting probability distribution of profit is a normal distribution in both cases, but in the 100bb case the middle of the distribution is higher, and the distribution is fatter (higher variance).*

This doesn't take into account the additional "room for pokering" with deeper stacks, which obviously is a large factor that this model doesn't account for, but it at least gives some concrete idea...


*you could also think of this as a random walk in 1-D with probability of moving right of .6 and probability of moving left of .4. In the case of larger step sizes the particle will move on average more quickly to the right, but it will also be able to more quickly move in either direction. same thing, just a more fun way to think about it.


hope thats ok