my SD dropped?

[Piz0wn0reD!!!!!!]

At 100NL full ring my SD was 36BB/100 (144$/hr). I just switched to SH and my SD droped to 25BB/100 (after 16,000 hands). Is this accurate? Also what does this mean? Is there LESS varience at the SH games? WTF?

Thanks

[gaming_mouse]

Yes,

I am fairly sure SH games have less variance (not 100% tho). This seems counter-inuitive since you will be playing more.

But, very very very roughly, you can think of it as follows: When you play a 10-handed game and enter a pot, you have a 1 in 10 chance of winning. In a 6-handed game, 1 in 6. It's like the difference between playing red/black in roulette versus playing a single numer -- except a less extreme version. And it's easy to see that the single number has higher variance.

Also jason1990, I think, was going to write an article on the accuracty of SD estimates -- I don't know if he did or not. That may have some bearing here, but I think the explanation above is the correct one, more or less. Especially given the size of your sample.

gm

[AngryCola]

I don't play much short-handed. But I thought I always heard the variance was higher. Everyone always recommends a larger bankroll for short-handed games, so I assumed the variance to be higher.

Seems logical that it would be higher. But I've been known to be wrong about these sorts of things.

[gaming_mouse]

I remembered where I read it. There's an article about it here:

http://games.cs.ualberta.ca/poker/IRC/mm-sh-var.txt

[Piz0wn0reD!!!!!!]

I also have heard from a lot of people that SH has bigger swings. If it does have less varience, thats great. In the first 16k hands, ive hardly run bad at all. I think ive only had 2 or 3 losing days. I better go knock on some wood, as im sure this trend wont last.

[gaming_mouse]

[ QUOTE ]
I also have heard from a lot of people that SH has bigger swings.

[/ QUOTE ]

Yeah, I have too. Perhaps this is one of those cases where common wisdom isn't correct. That article I posted makes a very strong case, backed up with a lot of evidence, for SH having less variance. I'm inclined to believe it barring other (non-anecdotal) evidence to the contrary.

gm

[uuDevil]

"They" also say that aggression increases variance, but there's plenty of evidence that the opposite is true. Although I don't play shorthanded, I expect the aggression level is greater than in full games. The combination of starting shorthanded with extra aggression should reduce the number of players seeing each street compared to full games. This means fewer hands will go to showdown and pots will be smaller. Small pots and a greater chance to win when you decide to play should give lower variance.

On the other hand, it is correct to play more loosely than in a full game. This would increase variance.

If someone could work out the details of these relationships (develop a simple model), it might make a good article for the online magazine.

[AngryCola]

[ QUOTE ]
If someone could work out the details of these relationships (develop a simple model), it might make a good article for the online magazine.

[/ QUOTE ]

I agree. There seems to be a lot of confusion about this issue.

I nominate gm to write the article. /images/graemlins/grin.gif

[gaming_mouse]

[ QUOTE ]
If someone could work out the details of these relationships (develop a simple model), it might make a good article for the online magazine.

[/ QUOTE ]

Trying to build a model here is the wrong way to approach the problem. All you need to do is analyze empirical data, which you cannot argue with. The article I linked to does that, and is overwhelming evidence that SH games have smaller variance.

The only objection anyone could have to have those results is that they are from the IRC database, rather than real money games. This is a fair objection, but I would be willing to bet the results will hold for real money as well.

Anyone with a large enough PT database can filter on 6 max games and see for themselves. I don't have enough hands to verify myself, but anyone else can. I'd like to see.

gm

EDIT: Holy sh*t, AngryCola. We literally repsonded at the same time to the exact same quote. Weird.

[uuDevil]

[ QUOTE ]
Trying to build a model here is the wrong way to approach the problem. All you need to do is analyze empirical data, which you cannot argue with. The article I linked to does that, and is overwhelming evidence that SH games have smaller variance.

[/ QUOTE ]
This is true if all you want to do is know whether variance is lesser or greater.

It would also be interesting to investigate why it is lesser or greater. What are the factors that influence variance? How relatively important are they? How do they interact? A model would help make this clear.

[AngryCola]

[ QUOTE ]
It would also be interesting to investigate why it is lesser or greater. What are the factors that influence variance? How relatively important are they? How do they interact? A model would help make this clear.

[/ QUOTE ]

:nods:

I don't know about anyone else, but I would find such an "article" to be very interesting.

[gaming_mouse]

[ QUOTE ]

This is true if all you want to do is know whether variance is lesser or greater.

It would also be interesting to investigate why it is lesser or greater.

[/ QUOTE ]

Yes. Fair enough.

[Izverg04]

[ QUOTE ]

I am fairly sure SH games have less variance (not 100% tho). This seems counter-inuitive since you will be playing more.

http://games.cs.ualberta.ca/poker/IRC/mm-sh-var.txt

That article I posted makes a very strong case, backed up with a lot of evidence, for SH having less variance. I'm inclined to believe it barring other (non-anecdotal) evidence to the contrary.


[/ QUOTE ]
Don't put too much value in that article. Just look at the numbers. I don't know what "IRC poker" is, but limit holdem doesn't have SD = 41 BB/100 hands for 10-handed and SD = 26 BB/100 hands for 6-handed play. Those numbers look like they've been taken from play-money VP$IP=100% tables. Only in that context these numbers make sense, since you'll have larger pots at full tables where everyone plays at least preflop.

It is nonsense to claim that short-handed limit hold'em has smaller variance than full-handed. I think anyone who plays both has numbers to back up the oppposite. Your VP$IP goes up quicker as the table gets shorter while the pot size get smaller slowly. This drives up the variance.

[AngryCola]

[ QUOTE ]
I think anyone who plays both has numbers to back up the oppposite.

[/ QUOTE ]

Then let them post their stats. Until then, all you are doing is guessing the other data is useless.

Plus, your other points have already been discussed.

[Izverg04]

I am not guessing that the data in the article is useless -- that should be clear.

I have only 5k hands of full tables (I play 6-handed), so personally I don't have enough stats.

Would you consider the polled SD numbers referenced in the other post as evidence? They indicate that 6-handed limit HE gives 10-15% larger SD than 10-handed.

[AngryCola]

[ QUOTE ]
I am not guessing that the data in the article is useless -- that should be clear.

[/ QUOTE ]

Honestly, I'm not smart enough to judge this. In these matters, I almost always defer to the likes of gaming mouse and BruceZ.

[Piz0wn0reD!!!!!!]

[ QUOTE ]
It is nonsense to claim that short-handed limit hold'em has smaller variance than full-handed. I think anyone who plays both has numbers to back up the oppposite. Your VP$IP goes up quicker as the table gets shorter while the pot size get smaller slowly. This drives up the variance.

[/ QUOTE ]

ok, this is what i thought. If this is true, how on earth is my SD 1/3 smaller in SH NL? or rather why?

[Izverg04]

Personally, I don't know because I don't play no limit. First of all, this could be a fluctuation. I estimate about 1.5 sigma between your numbers being equal with your samples.

I suppose that in no-limit variance is driven by how often you go all-in, not how often you voluntarily enter the pot. There is some positive correlation between the two for sure, and some negative correlation between the number of players and how often you go all-in. With more players, you find more often another strong hand to go all-in against your strong hand preflop. The same effect is not as important in limit because the pot never gets as big even with 2 strong hands against each other.

[gaming_mouse]

[ QUOTE ]
I am not guessing that the data in the article is useless -- that should be clear.

[/ QUOTE ]

It is not clear. It is only clear that it MAY not apply to real money games, and that is something I have already pointed out myself, if you read my other posts in this thread.

Moreover, the data that Piz0wn0reD posted in the other thread seems to support the evidence of the article, more or less. His friends limit SH SD is the same as his full ring SD, and that is for a very aggressive player.

So far, no one has posted ANY data that supports the common wisdom that SH games have a higher SD than full ring games. I am not claiming, and never have claimed, that the case for limit SH games having smaller SD has been proved 100% yet.

But until you, or someone else, posts data to the contrary you should not be saying that it's obvious that they have higher variance just because its accepted wisdom.

gm