vpFREE2 Forums

Cumulative tabulation of PDF challenge

#1

wrote:
> It has been reported that a program has been written to do this,

but I

> have yet to find it although I keep looking.

"nightoftheiguana2000" wrote

http://www.lotspiech.com/GamblersRuin.html

I do not see how the program in the above URL does that, Perhaps an
explanation in addition to the URL posting would help. Or someone
else wanting to be of help could give a more thorough explanation how
this is so. Sometimes what is obvious to someone is not obvious to
someone else without explanation. Or the posting may have been
misunderstood.

For reference purposes the challenge that I made is re-issued here
again:

I issue a challenge on forum for a willing person to make a
contribution to vp literature by constructing a tabulated cumulation
of the PDFs for various size sessions and for the popular games.
Someone else suggested to me privately that attention be paid to the
tail portions on both ends, due to the inability to discern small
changes in a graphical representation of same such as those shown on
Jazbo.com under video poker then under vp probabilities.

Some suggested tabulated acumulations of PDFs would be for say 2K
5K 10K and maybe 100K session sizes for the more popular games,
tabulated in a cumulative format so that a look up could be done
similar to using a Z table

For example one would enter the table for Session of 5000 hands of
JoB and be able to determine that 95 % of the time their outcome
would be X betting units or less. Or one could determine that
between A and B percent of the time they should make a profit of Y1
to Y2 betting units. To me this is the same kind of analysis that
Jazbo was showing in his cumulations of PDFs in graphical form.
Tabulation would increase the ability to be more precise.

DWK

···

--- In vpFREE@yahoogroups.com, "deuceswild1000" <deuceswild1000@y...>

#2

That's what the program does:

"The calculator tells you what your probability is of being in each
winnings range (including "retired" and "busted"), for the number of
hands shown. You enter your initial stake, and at what point you will
be willing to walk away a winner. (I hope this is not a completely
foreign concept for you.) You can run it for any number of hands."

wrote:

···

--- In vpFREE@yahoogroups.com, "deuceswild1000" <deuceswild1000@y...>

--- In vpFREE@yahoogroups.com, "deuceswild1000"
<deuceswild1000@y...>
> wrote:
> > It has been reported that a program has been written to do this,
but I
> > have yet to find it although I keep looking.

"nightoftheiguana2000" wrote

> http://www.lotspiech.com/GamblersRuin.html

I do not see how the program in the above URL does that, Perhaps an
explanation in addition to the URL posting would help. Or someone
else wanting to be of help could give a more thorough explanation how
this is so. Sometimes what is obvious to someone is not obvious to
someone else without explanation. Or the posting may have been
misunderstood.

For reference purposes the challenge that I made is re-issued here
again:

I issue a challenge on forum for a willing person to make a
contribution to vp literature by constructing a tabulated cumulation
of the PDFs for various size sessions and for the popular games.
Someone else suggested to me privately that attention be paid to the
tail portions on both ends, due to the inability to discern small
changes in a graphical representation of same such as those shown on
Jazbo.com under video poker then under vp probabilities.

Some suggested tabulated acumulations of PDFs would be for say 2K
5K 10K and maybe 100K session sizes for the more popular games,
tabulated in a cumulative format so that a look up could be done
similar to using a Z table

For example one would enter the table for Session of 5000 hands of
JoB and be able to determine that 95 % of the time their outcome
would be X betting units or less. Or one could determine that
between A and B percent of the time they should make a profit of Y1
to Y2 betting units. To me this is the same kind of analysis that
Jazbo was showing in his cumulations of PDFs in graphical form.
Tabulation would increase the ability to be more precise.

DWK

#3

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

That's what the program does:

"The calculator tells you what your probability is of being in each
winnings range (including "retired" and "busted"), for the number of
hands shown. You enter your initial stake, and at what point you will
be willing to walk away a winner. (I hope this is not a completely
foreign concept for you.) You can run it for any number of hands."

I understand that and have been familar with program for quite a few
years. What I am looking for is the final outcome of a given number
of hands regardless of win or loss. That program puts a constraint on
bankroll and or winnings/losses. Seems clear to me.

Also the program I was refering to is supposed to exist but has not
been releasd to the public the last I knew. .

#4

Ok, I volunteer to do this for JoB. I will produce the following
statistics:

mode, min (largest loss), max (largest win)
win/loss at 1%,2%,3%.... all the way up to say >95% or so (I haven't
decided how high yet, see note on the bottom)

I will do this for a range of number of hands:
1,10,100,1000,?,.....? (feedback would be welcome here)

I will also put the results in an excell table. Any other suggestions?

Unfortunately, I can't say for sure when I will be finished. BTW,I
was planning to save this stuff in a book I am writing which I may
never finish either-- oh well.

I guess the nice thing about doing is that evevyone will have the
data, and they can make thier own conclusions from it...

···

------------------------------
note: To speed things up, I will making 1 simplifing assumption.
Beyond some critical # of Royals, I will be stuffing all the
accumateled probability into 1 bin. Doing so ensures that the
integral of the PDF = 1. It also reduces the memory requirement
drastically. But it also causes the PDF to become somewhat inexact
(a better word would be integrated) after some critical probability.
Nonetheless, the effect is very small and can be computed. Conclusion,
the win/loss at the last probability bin (say 100%) will not mean
exactly the same thing as data for the other bins. No big deal. I
just thought I should disclose it though.

--- In vpFREE@yahoogroups.com, "deuceswild1000" <deuceswild1000@y...>
wrote:

What I am looking for is the final outcome of a given number
of hands regardless of win or loss. That program puts a constraint
on bankroll and or winnings/losses. Seems clear to me.

#5

suggestions?

First, let me say thanks.

Second, I would think that session sizes of less than say 2k would
not have much value. My orginal suggestions were 2K 5K 10K and 100K

Third, Please consider doing FPDW or NSUD someday, the reason is
probably obvious :slight_smile:

Fourth, I might have to break down and purchase Excell instead of just
using Works.

DWK deuceswild1000

···

--- In vpFREE@yahoogroups.com, "cdfsrule" <groups.yahoo@v...> wrote:

Ok, I volunteer to do this for JoB. I will produce the following
statistics:

mode, min (largest loss), max (largest win)
win/loss at 1%,2%,3%.... all the way up to say >95% or so (I haven't
decided how high yet, see note on the bottom)

I will do this for a range of number of hands:
1,10,100,1000,?,.....? (feedback would be welcome here)

I will also put the results in an excell table. Any other

#6

Since all those discussions were sparked by the diamond-in-a-day
bankroll problem, maybe using the session sizes that match
diamond-in-a-day would have some special significance as well.

JBQ

···

On 9/2/05, deuceswild1000 <deuceswild1000@yahoo.com> wrote:

Second, I would think that session sizes of less than say 2k would
not have much value. My orginal suggestions were 2K 5K 10K and 100K

#7

your welcome... but don't get too excited yet....

I took a deeper look at excell and found (remembered?) that, if I
activate the correct "add-in", that a Fast Fourier Transform (FFT)
function becomes available. Using the FFT, it is posible to
impliment a very fasy Convolution function, which is the basis of the
PDF computation (since I had forgoten that excel had an FFT function,
I described a manual process some emails ago, mathematically, the
process is known as a convolution). I also recall now (its funny how
my memory works) that Excel did some odd things with FFT's (like
internally not keeping the data in double precisions for FFT's or
something equally sinister). Anyhoo, rather than just give the PDF
results now, I am going to try set up an excell spreadsheet that will
let anyone compute the PDF for the game of their choice, including
any Wynn paytable that might pop up. This may take a while, since I
am NOT an excel expert. Moreover, though I find excel incredibly
useful, I hate using it!

--- In vpFREE@yahoogroups.com, "deuceswild1000" <deuceswild1000@y...>
wrote:

> Ok, I volunteer to do this for JoB. I will produce the following
> statistics:
>
> mode, min (largest loss), max (largest win)
> win/loss at 1%,2%,3%.... all the way up to say >95% or so (I

haven't

> decided how high yet, see note on the bottom)
>
> I will do this for a range of number of hands:
> 1,10,100,1000,?,.....? (feedback would be welcome here)
>
> I will also put the results in an excell table. Any other
suggestions?

First, let me say thanks.

Second, I would think that session sizes of less than say 2k would
not have much value. My orginal suggestions were 2K 5K 10K and

100K

Third, Please consider doing FPDW or NSUD someday, the reason is
probably obvious :slight_smile:

Fourth, I might have to break down and purchase Excell instead of

just

···

--- In vpFREE@yahoogroups.com, "cdfsrule" <groups.yahoo@v...> wrote:
using Works.

DWK deuceswild1000

#8

Well, I wasted the greater part of a perfectly good Saturday but I've
finished (a beta version). I will send it off to the administrator
for posting soon.

Here's what I've got so far: 1 excel spreadsheet. You supply the
paytable, the bet size, and number of hands and it computes the PDF.
There are still some improvements that are needed (the plotting sucks
for example) and I would appreciate your comments. BTW, it's being
shipped set up to go for JoB, using probabilities from the wizard of
odds.

Please keep in mind that this is a work in progress. In the future, I
hope to be able to add a RoR calculator also (for finite # of
hands). This will be an EXACT (within the limits of excels accuracy)
computation. BTW, you may already notice that there is a tiny
accuracy problem for very small probabilities (big losses for
example). This is an excel limitation, not a programming bug. Yet
one more reason to hate Microsoft.

I am making no warranty regarding this software. Use it at your own
risk.

···

----

Read me first:

Input fields are in yellow boxes with italic character formatting
Output fields are in blue (don't mess with them).
Also, don't mess with anything that isn't colored in. You will be
sorry if you do.

The code won't work if MACROs have been disabled. So enable Macros.

The code won't work unless you have installed the analysis add-in.
(You can do this under the tools menu)

To run, type cntrl-w. Excel will ask you 2x if you want to overwrite
the data that already exists. Click "ok". It won't hurt.
(Anyone know how to force Excel to stop this behavior?)

Always remember to type ctrl-w if after you make changes to the input
parameters.
You don't need to type ctrl-w if you just change the percentages in
the results box (on left)

Performance is not significantly effected by number of hands.
Performance is significantly affect by "Max # RF" input, so keep it
small. (You are warned). But if it is too small, the accuracy of the
computation will become worse for the highest probability levels.
This won't be significant until a very, very, very large number of
hands is reached.

One more thing.

I've put my name inside the Excel file (and in a GNU public license
statement located in the macro). Please don't go throwing it around
the list, especially if you want me to continue my development work.
Thanks.

First, let me say thanks.

Second, I would think that session sizes of less than say 2k would
not have much value. My orginal suggestions were 2K 5K 10K and

100K

Third, Please consider doing FPDW or NSUD someday, the reason is
probably obvious :slight_smile:

Fourth, I might have to break down and purchase Excell instead of

just

using Works.

DWK deuceswild1000

#9

Let me thank you for all this work! While not an
active participant inthe math discussions, I do read
them and look forward to trying out your spreadsheet.

Mark

···

--- cdfsrule <groups.yahoo@verizon.net> wrote:

Well, I wasted the greater part of a perfectly good
Saturday but I've
finished (a beta version). I will send it off to
the administrator
for posting soon.

Here's what I've got so far: 1 excel spreadsheet.
You supply the
paytable, the bet size, and number of hands and it
computes the PDF.
There are still some improvements that are needed
(the plotting sucks
for example) and I would appreciate your comments.
BTW, it's being
shipped set up to go for JoB, using probabilities
from the wizard of
odds.

Please keep in mind that this is a work in progress.
In the future, I
hope to be able to add a RoR calculator also (for
finite # of
hands). This will be an EXACT (within the limits of
excels accuracy)
computation. BTW, you may already notice that there
is a tiny
accuracy problem for very small probabilities (big
losses for
example). This is an excel limitation, not a
programming bug. Yet
one more reason to hate Microsoft.

I am making no warranty regarding this software.
Use it at your own
risk.

----

Read me first:

Input fields are in yellow boxes with italic
character formatting
Output fields are in blue (don't mess with them).
Also, don't mess with anything that isn't colored
in. You will be
sorry if you do.

The code won't work if MACROs have been disabled.
So enable Macros.

The code won't work unless you have installed the
analysis add-in.
(You can do this under the tools menu)

To run, type cntrl-w. Excel will ask you 2x if you
want to overwrite
the data that already exists. Click "ok". It won't
hurt.
(Anyone know how to force Excel to stop this
behavior?)

Always remember to type ctrl-w if after you make
changes to the input
parameters.
You don't need to type ctrl-w if you just change the
percentages in
the results box (on left)

Performance is not significantly effected by number
of hands.
Performance is significantly affect by "Max # RF"
input, so keep it
small. (You are warned). But if it is too small, the
accuracy of the
computation will become worse for the highest
probability levels.
This won't be significant until a very, very, very
large number of
hands is reached.

One more thing.

I've put my name inside the Excel file (and in a GNU
public license
statement located in the macro). Please don't go
throwing it around
the list, especially if you want me to continue my
development work.
Thanks.

> First, let me say thanks.
>
> Second, I would think that session sizes of less
than say 2k would
> not have much value. My orginal suggestions were
2K 5K 10K and
100K
>
> Third, Please consider doing FPDW or NSUD
someday, the reason is
> probably obvious :slight_smile:
>
> Fourth, I might have to break down and purchase
Excell instead of
just
> using Works.
>
> DWK deuceswild1000

____________________________________________________
Start your day with Yahoo! - make it your home page
http://www.yahoo.com/r/hs