vpFREE2 Forums

Stats Question - Variance

My question is math related, specifically how do you calculate
variance for a video poker payline?

The variance for a game is the sum of variances for each pay line.
Calculating the variance for a pay line is the piece I'm missing.

I have searched a few stats books that I own and the internet for a
formula to do this but have found only the following equations
regarding variance:

var = (x-u)^2 * p(x)
var = (x^2 * p(x)) - u^2

The examples given for variance calculations always seem to be based
on coin flipping or dice rolling and do not seem to work for video
poker. I would guess that's because these equations are used for
random events that have a normal distribution and video poker does not
have a normal distribution (or video poker is not random!).

If I use the above formula to calculate the variance of a royal flush
in a 9/6 Jacks+ game I get the following:

u = (493,512,264 * 800) / 19,933,230,517,200 = 0.0198 (I'm assuming
here that u = ev(x))
p(x) = 493,512,264 / 19,933,230,517,200 = 0.000024758
var = (800-0.0198)^2 * 0.000024758 = 15.84434

Every source I've found that states variance for paylines states the
variance for a royal flush as 15.8059.

I'm obviously missing something here, probably something very obvious
but I just can't seem to put my finger on it.

If anyone can help or point me to a source that explains it, I'd be
very grateful!

metal_guy_03842 wrote:

My question is math related, specifically how do you calculate
variance for a video poker payline?

The variance for a game is the sum of variances for each pay line.
Calculating the variance for a pay line is the piece I'm missing.

I have searched a few stats books that I own and the internet for a
formula to do this but have found only the following equations
regarding variance:

var = (x-u)^2 * p(x)
var = (x^2 * p(x)) - u^2

If I use the above formula to calculate the variance of a royal
flush in a 9/6 Jacks+ game I get the following:

u = (493,512,264 * 800) / 19,933,230,517,200 = 0.0198 (I'm assuming
here that u = ev(x))
p(x) = 493,512,264 / 19,933,230,517,200 = 0.000024758
var = (800-0.0198)^2 * 0.000024758 = 15.84434

Every source I've found that states variance for paylines states the
variance for a royal flush as 15.8059.

I'm obviously missing something here, probably something very
obvious but I just can't seem to put my finger on it.
If anyone can help or point me to a source that explains it, I'd be
very grateful!

It's a straightforward problem. "u" isn't the expected value of the
RF payout per hand, as you assume parenthetically. It's the overall
expected payout per hand ... in the case of Jacks, it's the game
return of 0.995439 bets. Substitute this value and you'll arrive at
the correct variance contribution to the game from the RF.

- Harry

Thank you! It works! A classic case of getting so locked into one way
of thinking you overlook the obvious.

ยทยทยท

--- In vpFREE@yahoogroups.com, "Harry Porter" <harry.porter@v...> wrote:

metal_guy_03842 wrote:
>
> My question is math related, specifically how do you calculate
> variance for a video poker payline?
>
> The variance for a game is the sum of variances for each pay line.
> Calculating the variance for a pay line is the piece I'm missing.
>
> I have searched a few stats books that I own and the internet for a
> formula to do this but have found only the following equations
> regarding variance:
>
> var = (x-u)^2 * p(x)
> var = (x^2 * p(x)) - u^2
>
> If I use the above formula to calculate the variance of a royal
> flush in a 9/6 Jacks+ game I get the following:
>
> u = (493,512,264 * 800) / 19,933,230,517,200 = 0.0198 (I'm assuming
> here that u = ev(x))
> p(x) = 493,512,264 / 19,933,230,517,200 = 0.000024758
> var = (800-0.0198)^2 * 0.000024758 = 15.84434
>
> Every source I've found that states variance for paylines states the
> variance for a royal flush as 15.8059.
>
> I'm obviously missing something here, probably something very
> obvious but I just can't seem to put my finger on it.
> If anyone can help or point me to a source that explains it, I'd be
> very grateful!

It's a straightforward problem. "u" isn't the expected value of the
RF payout per hand, as you assume parenthetically. It's the overall
expected payout per hand ... in the case of Jacks, it's the game
return of 0.995439 bets. Substitute this value and you'll arrive at
the correct variance contribution to the game from the RF.

- Harry