vpFREE2 Forums

Standard Deviation of multiple line play

#1

I have been following this conversation with interest. And I think the
conclusion that all the different formulas lead to the same final answer is true.

However be careful since the standard deviation is meant for a normal
distribution, which video poker is definately not. If you use the calculation
methods for a single deal on 50 play the answer will be a SD greater than what you
bet. I think that the SD is not too bad when enough hands are played, but
that it overestimates the loss. This is due to the variance being quite a bit
higher due to the royal flush. If the variance was calculated for returns
without the royal flush ( which I think can be done) the variance would be
somewhat lower.

I would appreciate any comments on this possibility.

Thanks

Bob From Erie

[Non-text portions of this message have been removed]

#2

You are correct.
You should discount wins that have an average cycle greater than the
number of hands you are looking at. For example: 4900 hands of fpdw,
that's less than the deuces cycle of 4909 hands and less than the
royal cycle of 45282, so subtract out their contributions to ER and Var:
new ER=1.007619612-0.017667091-0.04074064=0.949211881 (~-5%)
new Var=25.83461805-14.0980921-8.066232646=3.670293306
for multiplays you'd have to calculate a new covariance, my guess is
it would be at least one half, possibly a third of the original value

I have been following this conversation with interest. And I think the
conclusion that all the different formulas lead to the same final

answer is true.

However be careful since the standard deviation is meant for a normal
distribution, which video poker is definately not. If you use the

calculation

methods for a single deal on 50 play the answer will be a SD

greater than what you

bet. I think that the SD is not too bad when enough hands are

played, but

that it overestimates the loss. This is due to the variance being

quite a bit

higher due to the royal flush. If the variance was calculated for

returns

without the royal flush ( which I think can be done) the variance

would be

···

--- In vpFREE@yahoogroups.com, BOBBJ@a... wrote:

somewhat lower.

I would appreciate any comments on this possibility.

Thanks

Bob From Erie

[Non-text portions of this message have been removed]

#3

Dear NOTI,

Hi! Again, I am just seeking information and understanding.

In the second formula, below, what does 14.0980921 represent and what
does 8.066232646 represent, in your computation of the "new" variance?

Just for kicks, I ran the "Analyze Game" for FPDW in WinPoker, setting the
"return" for the RF and 4D to zero. It gave an ER of 95.4491% and a Game
Variance (in Single Coin Bets) of 3.73505.

I am assuming that the "differences" with the WinPoker numbers and yours
are the result of the fact that the "analysis" is performed, playing a different
strategy, not seeking the RF or 4D.

.....bl

--- In vpFREE@yahoogroups.com, "nightoftheiguana2000" <
nightoftheiguana2000@y...> wrote:

···

You are correct.
You should discount wins that have an average cycle greater than the
number of hands you are looking at. For example: 4900 hands of fpdw,
that's less than the deuces cycle of 4909 hands and less than the
royal cycle of 45282, so subtract out their contributions to ER and Var:
new ER=1.007619612-0.017667091-0.04074064=0.949211881 (~-5%)
new Var=25.83461805-14.0980921-8.066232646=3.670293306
for multiplays you'd have to calculate a new covariance, my guess is
it would be at least one half, possibly a third of the original value

#4

Dear NOTI,

Hi! Again, I am just seeking information and understanding.

In the second formula, below, what does 14.0980921 represent and what
does 8.066232646 represent, in your computation of the "new" variance?

~14 is the variance of a royal flush, ~8 is the variance of deuces

Just for kicks, I ran the "Analyze Game" for FPDW in WinPoker,

setting the

"return" for the RF and 4D to zero. It gave an ER of 95.4491% and a

Game

Variance (in Single Coin Bets) of 3.73505.

I am assuming that the "differences" with the WinPoker numbers and

yours

are the result of the fact that the "analysis" is performed, playing

a different

strategy, not seeking the RF or 4D.

Correct

.....bl

--- In vpFREE@yahoogroups.com, "nightoftheiguana2000" <
nightoftheiguana2000@y...> wrote:
> You are correct.
> You should discount wins that have an average cycle greater than the
> number of hands you are looking at. For example: 4900 hands of fpdw,
> that's less than the deuces cycle of 4909 hands and less than the
> royal cycle of 45282, so subtract out their contributions to ER

and Var:

···

--- In vpFREE@yahoogroups.com, "bornloser1537" <bornloser1537@y...> wrote:

> new ER=1.007619612-0.017667091-0.04074064=0.949211881 (~-5%)
> new Var=25.83461805-14.0980921-8.066232646=3.670293306
> for multiplays you'd have to calculate a new covariance, my guess is
> it would be at least one half, possibly a third of the original value
>