anyone know how to calculate standard deviation?

[HonchoOverload]

I know the formula, I'm just not quite sure what to plug in to make it relevant to poker stats.

[bluefeet]

http://kimsal.com/rabbit_pancake.jpg

If it's to check fun things like OOTM/ITM runs, ROI swings and the like. Check out rvg72's
handy simulator thingy (http://forumserver.twoplustwo.com/showthreaded.php?Cat=0&Number=3218857&page=0)

[citanul]

[ QUOTE ]
I know the formula, I'm just not quite sure what to plug in to make it relevant to poker stats.

[/ QUOTE ]

i have no pictures of bunnies.

to find the standard deviation of something, you plug in the numbers you would expect. to make it "relevant" to poker stats, you plug in the numbers for that stat. it's mostly worthless though. you should really, really, really not worry about it, and if you are going to worry about it, get a free tool that does all your stats including teh stupid ones.

c

[HonchoOverload]

Let me tell you what I'm trying to do. I need to calculate the SD because all the automated tools deal with 9 handed/10 handed sngs, and I play 6 handed where only 2 get paid. I want to know how much 3 less people cuts down the variance.

[splashpot]

[ QUOTE ]
Let me tell you what I'm trying to do. I need to calculate the SD because all the automated tools deal with 9 handed/10 handed sngs, and I play 6 handed where only 2 get paid. I want to know how much 3 less people cuts down the variance.

[/ QUOTE ]
It's still 33% get paid. I don't see how it would change at all from 9 man games.

Not taking into consideration difference in 1-2 payout.

[HonchoOverload]

[ QUOTE ]
[ QUOTE ]
Let me tell you what I'm trying to do. I need to calculate the SD because all the automated tools deal with 9 handed/10 handed sngs, and I play 6 handed where only 2 get paid. I want to know how much 3 less people cuts down the variance.

[/ QUOTE ]
It's still 33% get paid. I don't see how it would change at all from 9 man games.

Not taking into consideration difference in 1-2 payout.

[/ QUOTE ]

Yeah the payout is the same, but the variance is different and I want to know how much, because I want to know how many tourneys give me a relevant data set.

In the FAQ, one of the mods stated that 500-1000 tourneys is good in terms of predicting a baseline for ROI and such. Is the same true for 6-handed? I would think it would be less because less people playing means less variance....can I just chop it by a third or is it more complicated?

Also, in the example in the FAQ the mod threw out an SD of 19. Is that arbitrary?

[Shillx]

Post deleted by Shillx

[HonchoOverload]

[ QUOTE ]
Okay lets say that you are a losing player. You finish in all places in equal amounts and your ROI is -9.09%. Your SD would be 1.38 buyins/game and that is lower then it would be as a -9.09% ROI in a 10 man SNG. If you finish in each spot equally in a full SNG, your SD would be 1.52 buyins/game.

[/ QUOTE ]

Thanks, but can you walk me through exactly how you calculated that?

[splashpot]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
Let me tell you what I'm trying to do. I need to calculate the SD because all the automated tools deal with 9 handed/10 handed sngs, and I play 6 handed where only 2 get paid. I want to know how much 3 less people cuts down the variance.

[/ QUOTE ]
It's still 33% get paid. I don't see how it would change at all from 9 man games.

Not taking into consideration difference in 1-2 payout.

[/ QUOTE ]

Yeah the payout is the same, but the variance is different and I want to know how much, because I want to know how many tourneys give me a relevant data set.

In the FAQ, one of the mods stated that 500-1000 tourneys is good in terms of predicting a baseline for ROI and such. Is the same true for 6-handed? I would think it would be less because less people playing means less variance....can I just chop it by a third or is it more complicated?

Also, in the example in the FAQ the mod threw out an SD of 19. Is that arbitrary?

[/ QUOTE ]
Why would the variance be different if you get ITM the same % of the time? Why does fewer people mean lower variance?

[HonchoOverload]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
Let me tell you what I'm trying to do. I need to calculate the SD because all the automated tools deal with 9 handed/10 handed sngs, and I play 6 handed where only 2 get paid. I want to know how much 3 less people cuts down the variance.

[/ QUOTE ]
It's still 33% get paid. I don't see how it would change at all from 9 man games.

Not taking into consideration difference in 1-2 payout.

[/ QUOTE ]

Yeah the payout is the same, but the variance is different and I want to know how much, because I want to know how many tourneys give me a relevant data set.

In the FAQ, one of the mods stated that 500-1000 tourneys is good in terms of predicting a baseline for ROI and such. Is the same true for 6-handed? I would think it would be less because less people playing means less variance....can I just chop it by a third or is it more complicated?

Also, in the example in the FAQ the mod threw out an SD of 19. Is that arbitrary?

[/ QUOTE ]
Why would the variance be different if you get ITM the same % of the time? Why does fewer people mean lower variance?

[/ QUOTE ]

Variance has nothing to do with % of people getting paid out. Since there are three more places I can finish in, that jacks up the variance. More possibilities, more variance.

[splashpot]

[ QUOTE ]
Variance has nothing to do with % of people getting paid out. Since there are three more places I can finish in, that jacks up the variance. More possibilities, more variance.

[/ QUOTE ]
What? This is totally wrong. If there was a tournament that paid 10 places, but only had 11 entrants, the variance would be extremely low. More places you can finish.

[Shillx]

Heh I had it all right but I got rid of it.

For 6 max here would be your numbers (assume a 22 buyin)....

-22, -22, -22, -22, +18, +58

xbar = - $2
s = root ((4*20*20 + 20*20 + 60*60)/6) = $28.28 = 1.38 buy-ins

For 10 man SNGs

-22, -22, -22, -22, -22, -22, -22, +18, +38, +78

xbar = - $2
s = root ((7*400 + 400 + 1600 + 6400)/10)) = $33.46 = 1.52 buy-ins

Brad

Edit - The better the player you are, the more your varience goes up. So even if you are a losing player, you have quite a bit of varience in these games. It would not be uncommon for a good player to have a varience of 1.6 BI or higher for regualr 10 man SNGs. If you sucked ass (you finished OOTM everytime) your SD would be nil.

Re-edit: The SD assumes that you have played an infinate # of SNGs. Someone who has only played 6 SNGs with these results will have an SD square(6/5) larger then the number I listed.

[HonchoOverload]

[ QUOTE ]
[ QUOTE ]
Variance has nothing to do with % of people getting paid out. Since there are three more places I can finish in, that jacks up the variance. More possibilities, more variance.

[/ QUOTE ]
What? This is totally wrong. If there was a tournament that paid 10 places, but only had 11 entrants, the variance would be extremely low. More places you can finish.

[/ QUOTE ]

you're looking at this in the wrong way. If I have 100,000 tournaments of data of 6-table and 9-table tournaments for the same player, I guarantee you that the Standard Deviation and the variance itself (as the formula above proves) will be greater for 9-player tournaments.

Thanks shillx!

[splashpot]

Correct me if I'm wrong, but ITM% and payout% is what changes SD, not the number of people. For example, a 20 man SNG that paid 25% top 2, 15% 3-4, and 10% 5-6 would have the exact same SD as the normal 10 man party SNGs.

[splashpot]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
Variance has nothing to do with % of people getting paid out. Since there are three more places I can finish in, that jacks up the variance. More possibilities, more variance.

[/ QUOTE ]
What? This is totally wrong. If there was a tournament that paid 10 places, but only had 11 entrants, the variance would be extremely low. More places you can finish.

[/ QUOTE ]

you're looking at this in the wrong way. If I have 100,000 tournaments of data of 6-table and 9-table tournaments for the same player, I guarantee you that the Standard Deviation and the variance itself (as the formula above proves) will be greater for 9-player tournaments.

Thanks shillx!

[/ QUOTE ]
The reason it's different is becuase 9 man SNGs have 50/30/20 payout and the 6 man ones have 65/35 payout. Not beacuse there are different amount of entrants.

[Shillx]

Let's say that there is a 100 man lotto. Entry is $1 and winner takes all. What is your SD in terms of buy-in?

xbar = 0

s = root ((100 + 100*100)/100) = $10.05 or 10.05 buy-ins

As you can see, a big top prize creates big varience. This is why MTTs have such huge varience (provided that you have a shot at winning). If they had a 10 man SNG where winner takes all (assume it is a $22), it would have much more varience then a typical Party 22. You will win 10% of the time.

ROI = -9.09%

s = root ((3600 + 32400)/10) = $60 = 2.73 buy-ins

[splashpot]

[ QUOTE ]
Let's say that there is a 100 man lotto. Entry is $1 and winner takes all. What is your SD in terms of buy-in?

xbar = 0

s = root ((100 + 100*100)/100) = $10.05 or 10.05 buy-ins

As you can see, a big top prize creates big varience. This is why MTTs have such huge varience (provided that you have a shot at winning). If they had a 10 man SNG where winner takes all (assume it is a $22), it would have much more varience then a typical Party 22. You will win 10% of the time.

ROI = -9.09%

s = root ((3600 + 32400)/10) = $60 = 2.73 buy-ins

[/ QUOTE ]
I don't understand where you're going with this. Yea, that all makes sense, but it has nothing to do with what I said. Do you agree or disagree with me?

[Shillx]

Yeah you are exactly right with what you said

[HonchoOverload]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
Variance has nothing to do with % of people getting paid out. Since there are three more places I can finish in, that jacks up the variance. More possibilities, more variance.

[/ QUOTE ]
What? This is totally wrong. If there was a tournament that paid 10 places, but only had 11 entrants, the variance would be extremely low. More places you can finish.

[/ QUOTE ]

you're looking at this in the wrong way. If I have 100,000 tournaments of data of 6-table and 9-table tournaments for the same player, I guarantee you that the Standard Deviation and the variance itself (as the formula above proves) will be greater for 9-player tournaments.

Thanks shillx!

[/ QUOTE ]
The reason it's different is becuase 9 man SNGs have 50/30/20 payout and the 6 man ones have 65/35 payout. Not beacuse there are different amount of entrants.

[/ QUOTE ]

You may be right. My head sure hurts.