A (probably) meaningless statistic Re WinPoker

[jackessiebabe]

I have had BDWP6 in my personal laptop since it was published.
Prior to that I had WP5, but when I replaced it with the later
version, I neglected to keep the stats.
As of last night, I played 1,113,520 hands on this program. Since I
always play multiline (3 hands at time)in order to keep from getting
bored, this amounts to 3,340,860 hands played.*
During this period of time, I've snagged 88 Royals, among the
various games I play. This amounts to approximately one RF every
37,964 hands. Pretty close to the long term EV, wouldn't you agree?
BTW, I also play with "warn" on, and never overide the expert
strategy, so this is a result of 100% perfect play.
FWIW, I play only what I actually play in Nevada casinos (FPDW,
FPKBJW, 10/7DB, 9/6DDB, and two games on which I changed the
payoffs, Downtown Deuces and 8/5Job/$3000prog.)

This may all be meaningless, but, as alawys, I welcome any comments.

               Babe
P.S. *When you play 3,4, or 10 hands at a time, the program only
registers 1 hand played. Does anyone know why?

[blaw57]

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

FWIW, I play only what I actually play in Nevada casinos (FPDW,
FPKBJW, 10/7DB, 9/6DDB, and two games on which I changed the
payoffs, Downtown Deuces and 8/5Job/$3000prog.)

This may all be meaningless, but, as alawys, I welcome any comments.

               Babe

Not meaningless at all, but very interesting. In fact, it bears on
the recent debate about non-randomness. I assume you've never reset
the counters, and that you're reading the stats you quoted out of the
program itself? If so, that would mean you can give a breakdown by
game, and we can see how far off perfect play would be after 1
million hands.

Can you give us the # of hands, # royals and W/L by game played? What
I want to do (or you can do in Winpoker/Excel if you are so inclined)
is to figure for each game how close to normal the results are for
each game for separate groups of royal and non-royal hands.

Of course, your results may be (and I guess, statistically speaking,
probably are) solidly in the middle of the famed bell curve, and that
would not disprove the possibility of extremes existing elsewhere.
But if you happen to some anomalies in the play, this may be the
perfect test case to illustrate normal statistical fluctuations, even
after a large number of hands have been played. There is no question
about record-keeping accuracy, no question about whether perfect
strategy was use, no question about whether the game chips had been
manipulated.

[jackessiebabe]

--- In vpFREE@yahoogroups.com, "jackessiebabe"

<jackessiebabe@y...>

wrote:

> FWIW, I play only what I actually play in Nevada casinos (FPDW,
> FPKBJW, 10/7DB, 9/6DDB, and two games on which I changed the
> payoffs, Downtown Deuces and 8/5Job/$3000prog.)
>
> This may all be meaningless, but, as alawys, I welcome any

comments.

>
> Babe

Not meaningless at all, but very interesting. In fact, it bears on
the recent debate about non-randomness. I assume you've never

reset

the counters, and that you're reading the stats you quoted out of

the

program itself? If so, that would mean you can give a breakdown by
game, and we can see how far off perfect play would be after 1
million hands.

Can you give us the # of hands, # royals and W/L by game played?

What

I want to do (or you can do in Winpoker/Excel if you are so

inclined)

is to figure for each game how close to normal the results are for
each game for separate groups of royal and non-royal hands.

Of course, your results may be (and I guess, statistically

speaking,

probably are) solidly in the middle of the famed bell curve, and

that

would not disprove the possibility of extremes existing elsewhere.
But if you happen to some anomalies in the play, this may be the
perfect test case to illustrate normal statistical fluctuations,

even

after a large number of hands have been played. There is no

question

about record-keeping accuracy, no question about whether perfect
strategy was use, no question about whether the game chips had

been

···

--- In vpFREE@yahoogroups.com, "blaw57" <blaw57@y...> wrote:

manipulated.

++++++++++++++++++++++++++++++++++++++++++++++

Hi blaw57

Your wish is my command. I do indeed have the info you seek in my
laptop.

AS soon as I leave the office, I shall garner these stats and post
them, either tonight or tomorrow.

Best, Babe

[brucem]

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

Your wish is my command. I do indeed have the info you seek in my
laptop.

AS soon as I leave the office, I shall garner these stats and post
them, either tonight or tomorrow.

Best, Babe

I'm interests me as well. I definitely think it becomes normalized,
and the return you experienced should be very close to the expected
value after 1,000,000+ games played. There is one issue,
however, that bothers me a tad, although its not supposed to happen.
That is, is the RNG in BDPW
identical functionally to the one in a casino VP machine?
Personally, I think they are identical.

[rgmustain]

There is one issue,

however, that bothers me a tad, although its not supposed to

happen.

That is, is the RNG in BDPW
identical functionally to the one in a casino VP machine?
Personally, I think they are identical.

Depends on what you mean by "functionally". While they may be
equivalent from a randomness perspectitive, they are definitely not
identical.

Dick

···

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

[jackessiebabe]

--- In vpFREE@yahoogroups.com, "jackessiebabe"

<jackessiebabe@y...>

wrote:

> FWIW, I play only what I actually play in Nevada casinos (FPDW,
> FPKBJW, 10/7DB, 9/6DDB, and two games on which I changed the
> payoffs, Downtown Deuces and 8/5Job/$3000prog.)
>
> This may all be meaningless, but, as alawys, I welcome any

comments.

>
> Babe

Not meaningless at all, but very interesting. In fact, it bears on
the recent debate about non-randomness. I assume you've never

reset

the counters, and that you're reading the stats you quoted out of

the

···

--- In vpFREE@yahoogroups.com, "blaw57" <blaw57@y...> wrote:

program itself?

++++++++++++++++++++++++++++++++++++

That is correct. I have never re-set the counter and the info below
are the stats taken directly from the BDWP program.

+++++++++++++++++++++++++++++++++++++++++++

If so, that would mean you can give a breakdown by

game, and we can see how far off perfect play would be after 1
million hands.

Can you give us the # of hands, # royals and W/L by game played? W

++++++++++++++++++++++++++++++++++++++++++++

Please see program stats below:

Game EV # of Hands Played* # of RF Actual Payback %

8/5BP 99.16 66,164 5 100.34
                   (198,492)*

9/6BPD 99.64 34,016 0 98.46
                   (102,048)*

9/6DDB 98.98 230,004 21 99.16
                   (690,012)*

9/6DJPT 100.34 55,557 6 100.20
                   (166,671)*

10/7DB 100.17 143,778 10 100.98
                   (431,133)*

FPDW 100.76 111,904 5 100.33
                   (335,712)*

FPKBJW 100.64 290,592 30** 100.42
                   (871,776)*
Downtown
DW 100.92 51,920 3 101.30
                   (155,160)*

8/5JoB
$3200Prog 102.30 101,715 6** 101.81
                   (305,145)*

DWBonusP 99.45 28,467 2 99.95
                    (85,401)*

Total "Draws": 1,114,117

Total "Hands": 3,342,351*

Total Royal Flushes: 88

Average # of Hands played to snag a RF: 37,981

*Top figure is # of draws. Bottom (parenthetical) figure is actual
number of hands played at triple-play.
I hope this info is of some interest to Blaw, Brumar and whoever
cares about this kind of math.

[jackessiebabe]

Not meaningless at all, but very interesting. In fact, it bears on
the recent debate about non-randomness. I assume you've never

reset

the counters, and that you're reading the stats you quoted out of

the

···

--- In vpFREE@yahoogroups.com, "blaw57" <blaw57@y...> wrote:

program itself? If so, that would mean you can give a breakdown by
game, and we can see how far off perfect play would be after 1
million hands.

+++++++++++++++++++++++++++

These are the raw, unadulerated stats from the

program. No counters were ever reset on this laptop.

+++++++++++++++++++++++++++

Can you give us the # of hands, # royals and W/L by game played?

+++++++++++++++++++++++++++

Please see the following "spreadsheet":

                 <snip>

Game EV Hands played* # of RF Actual Payback %

8/5BP 99.16 66,164 5 100.34
                      (198,492)*

9/6BPDXL 99.64 34,016 0 98.46
                      (102,048)*

9/6DDB 98.98 230,004 21 99.16
                      (690,012)*

9/6DDJPOT 100.34 55,557 6 100.98
                      (166,671)*

10/7DB 100.17 143,778 10 100.98
                      (431,133)*

FPDW 100,76 111,904 5 100.33
                       (335,712)*

FPKBJW 100.64 290,592 30** 100.42
                       (871,776)*

DWNTWNDW 100.92 51,920 3 101.30
                       (155,160)*
           
8/5JOB 102.30 101,715 6** 101.81
$3200PROG (305,145)*

BonusDW 99.45 28,467 2 99.95
                        (85,401)*

Total # of "Draws": 1,114,117

Total # of "Hands": 3,342,351

Total # of Royals: 88

* Figures in Parenthesis are ACTUAL # of Hands PLayed at triple play

**Some triple play included multiple RF on same draw.

I hope this is of some interest to a few of you at least. As usual,
I welcome any feedback, positive or otherwise.

                      Babe

[brucem]

::::::::::::
Thats what I meant by identical functionality ... that they both
produce a random result but using different methods. Actually its
more than the RNG, its the mapping method based on the random numbers
that also probably differs. Do you know how the RNG and mapping
scheme works in casino VP machines? I have never seen a detail
explanation ... just a very general description.

···

--- In vpFREE@yahoogroups.com, "rgmustain" <rgmustain@a...> wrote:

--- In vpFREE@yahoogroups.com, "brumar_lv" <brumar_lv@y...> wrote:
  There is one issue,
> however, that bothers me a tad, although its not supposed to
happen.
> That is, is the RNG in BDPW
> identical functionally to the one in a casino VP machine?
> Personally, I think they are identical.

Depends on what you mean by "functionally". While they may be
equivalent from a randomness perspectitive, they are definitely not
identical.

Dick

[rgmustain]

>
> There is one issue,
> > however, that bothers me a tad, although its not supposed to
> happen.
> > That is, is the RNG in BDPW
> > identical functionally to the one in a casino VP machine?
> > Personally, I think they are identical.
>
> Depends on what you mean by "functionally". While they may be
> equivalent from a randomness perspectitive, they are definitely

not

> identical.
>
> Dick
::::::::::::
Thats what I meant by identical functionality ... that they both
produce a random result but using different methods. Actually its
more than the RNG, its the mapping method based on the random

numbers

that also probably differs.

Probably true. There are many possible mapping methods.

Do you know how the RNG and mapping
scheme works in casino VP machines? I have never seen a detail
explanation ... just a very general description.

Me either. The easiest method I can think of offhand would be to
simply divide the random number by 52 and use the remainder as an
index into a fixed card table. Then, get the next random number and
repeat (discarding dups if they appear) until five cards are
selected. Same for the draw. Since the PRNG would run continually
between deal, draw and next deal, I can't think of any problems with
this simple approach. Of course, that doesn't mean there isn't one.

Dick

···

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

--- In vpFREE@yahoogroups.com, "rgmustain" <rgmustain@a...> wrote:
> --- In vpFREE@yahoogroups.com, "brumar_lv" <brumar_lv@y...> wrote:

[Harry_Porter]

rgmustain wrote:

Me either. The easiest method I can think of offhand would be to
simply divide the random number by 52 and use the remainder as an
index into a fixed card table. Then, get the next random number and
repeat (discarding dups if they appear) until five cards are
selected. Same for the draw. Since the PRNG would run continually
between deal, draw and next deal, I can't think of any problems with
this simple approach. Of course, that doesn't mean there isn't one.

Speculating (quite a bit), I'd expect that if newer generation
machines had a 32-bit processor, it would be more efficient to do an
algorithmic mapping directly to each of the 312 mil. possible ordered
hands - generating the hand in a single pass. By "algorithmic" I'm
suggesting that there wouldn't be a physical map in the program per se
(e.g. 1 = ac 2c 3c 4c 5c), but that a fairly straightforward
calculation would generate the appropriate 5 card hand from the random
variable.

However, your suggested method seems well suited for older generation
machines and it's quite possible that the method would simply be
carried forward to newer machines unaltered since efficiency isn't
critical in a vp machine.

As far as winpoker is concerned, I'd expect the simplicity of your
suggested method would make it very likely the method used.

The subject is, of course, totally academic. But it would be
fascination is Jim Wolf were to weigh in with the method used in
Frugal VP (and any knowledge he might have about that applied in
machines in the casino.

- Harry

[rgmustain]

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

rgmustain wrote:
> Me either. The easiest method I can think of offhand would be to
> simply divide the random number by 52 and use the remainder as an
> index into a fixed card table. Then, get the next random number

and

> repeat (discarding dups if they appear) until five cards are
> selected. Same for the draw. Since the PRNG would run continually
> between deal, draw and next deal, I can't think of any problems

with

> this simple approach. Of course, that doesn't mean there isn't

one.

Speculating (quite a bit), I'd expect that if newer generation
machines had a 32-bit processor, it would be more efficient to do an
algorithmic mapping directly to each of the 312 mil. possible

ordered

hands - generating the hand in a single pass. By "algorithmic" I'm
suggesting that there wouldn't be a physical map in the program per

se

(e.g. 1 = ac 2c 3c 4c 5c), but that a fairly straightforward
calculation would generate the appropriate 5 card hand from the

random

variable.

Since 2**32 is greater than 312 million this may be possible. And it
would also be possible to make the PRNG 64 bits or greater which
could be mapped more easily. As I mentioned, there are a lot of
potential mappings.

However, your suggested method seems well suited for older

generation

machines and it's quite possible that the method would simply be
carried forward to newer machines unaltered since efficiency isn't
critical in a vp machine.

As far as winpoker is concerned, I'd expect the simplicity of your
suggested method would make it very likely the method used.

The subject is, of course, totally academic. But it would be
fascination is Jim Wolf were to weigh in with the method used in
Frugal VP (and any knowledge he might have about that applied in
machines in the casino.

Academic? Ya think?

Of course, that never stops us from contemplating ... There may be
other considerations that the developers of the program had to
consider that we are ignoring. It's possible the gaming commission
placed some requirements on the program. E.g. that a shuffled virtual
deck must be demonstated in the program. Also, they could be a little
paranoid about someone discovering the algorithm and modifying it, so
they intentionally made it more complex.

Dick

[vp_nbi]

rgmustain wrote:
> Me either. The easiest method I can think of offhand would be to
> simply divide the random number by 52 and use the remainder as an
> index into a fixed card table. Then, get the next random number and
> repeat (discarding dups if they appear) until five cards are
> selected. Same for the draw. Since the PRNG would run continually
> between deal, draw and next deal, I can't think of any problems with
> this simple approach. Of course, that doesn't mean there isn't one.

Speculating (quite a bit), I'd expect that if newer generation
machines had a 32-bit processor, it would be more efficient to do an
algorithmic mapping directly to each of the 312 mil. possible ordered
hands - generating the hand in a single pass. By "algorithmic" I'm
suggesting that there wouldn't be a physical map in the program per se
(e.g. 1 = ac 2c 3c 4c 5c), but that a fairly straightforward
calculation would generate the appropriate 5 card hand from the random
variable.

The best RNG is hardware based. VIA has announced a chipset that
supports RNG based on electrical noise. Supposedly Intel has some new
generation chips that yield RNG. Also there are PC plugin devices that
yield random numbers based on thermal noise:
http://www.protego.se/sg100_en.htm#PricesandAvailability

Newer generation machines could utilize any of these which are
superior to computational RNG, but in the absence of regulatory
pressure it seems unlikely that vendors would spend the extra
manufacturing dollars. I suppose hardware based RNG could be a
competitive marketing point, but I'm not holding my breath waiting for
a manufacturer to lead in this area.

···

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

However, your suggested method seems well suited for older generation
machines and it's quite possible that the method would simply be
carried forward to newer machines unaltered since efficiency isn't
critical in a vp machine.

As far as winpoker is concerned, I'd expect the simplicity of your
suggested method would make it very likely the method used.

The subject is, of course, totally academic. But it would be
fascination is Jim Wolf were to weigh in with the method used in
Frugal VP (and any knowledge he might have about that applied in
machines in the casino.

- Harry