Hello,
I am considering some short term play above my normal limits to qualify
for some promotions, but I would like some help calculating the range
of my exposure to loss. For example, playing 9/6 JoB at $1, is there a
calculator that can tell me the range of possible outcomes for a given
number of hands? Say I play 200 hands, what does the distribution of
possible net results look like?
Thanks!
Danton wrote:
I am considering some short term play above my normal limits to
qualify for some promotions, but I would like some help calculating
the range of my exposure to loss. For example, playing 9/6 JoB at $1,
is there a calculator that can tell me the range of possible outcomes
for a given number of hands? Say I play 200 hands, what does the
distribution of possible net results look like?
Two products fit the bill:
- Dunbar's Risk Analyzer for Video Poker (DRA-VP)
- Video Poker for Winners (VPW)
DRA-VP is spreadsheet based (Excel) and thus requires that you have
access to Excel. It's a very fine product and will directly address
your need, even accommodating cashback, if you wish. It also does
long-term ROR/bankroll analysis.
VPW is a broad featured tutor (that goes well beyond Bob Dancer's
previously endorsed product, Winpoker). It includes a bankroll
analysis module that is comparable to that in DRA. It doesn't
accommodate cashback in the analysis; a very modest disadvantage.
(Frankly, I expect players generally don't make allowance for expected
cb in setting a "quit" point -- aside from that, you can easily
manually adjust net results.)
I think the DRA-VP analysis is a bit cleaner inasmuch as it expresses
the outcome distribution both graphically and in tabular form. VPW
only provides a graphic and you must pull numbers from the graph if
you want to construct a table -- something I desire each run and find
an awkward manual task.
However, the spreadsheet format of DRA-VP is a little more awkward to
work with than VPW and VPW has very strong appeal in being a broad
featured program that extends beyond just bankroll analysis. One key
advantage of VPW is that it accommodates multiline programs --
standard n-play, plus MultiStrike and Super Times Pay.
- Harry
There's a simple risk of ruin calculator here:
http://www.lotspiech.com/poker/GamblersRuin.html
Its based on quarters.
----- Original Message -----
From: Danton
To: vpFREE@yahoogroups.com
Sent: Wednesday, May 16, 2007 6:46 PM
Subject: [vpFREE] Calculation of Risk of Loss
Hello,
I am considering some short term play above my normal limits to qualify
for some promotions, but I would like some help calculating the range
of my exposure to loss. For example, playing 9/6 JoB at $1, is there a
calculator that can tell me the range of possible outcomes for a given
number of hands? Say I play 200 hands, what does the distribution of
possible net results look like?
Thanks!
[Non-text portions of this message have been removed]
--- In vpFREE@yahoogroups.com, "Harry Porter" <harry.porter@...>
wrote:
Danton wrote:
> I am considering some short term play above my normal limits to
> qualify for some promotions, but I would like some help
calculating
> the range of my exposure to loss. For example, playing 9/6 JoB at
$1,
> is there a calculator that can tell me the range of possible
outcomes
> for a given number of hands? Say I play 200 hands, what does the
> distribution of possible net results look like?Two products fit the bill:
- Dunbar's Risk Analyzer for Video Poker (DRA-VP)
- Video Poker for Winners (VPW)
A few questions:
Both of these will give me a graphical representation of the outcomes
of a given trip stake,using a given number of hands played during the
trip (I assume either you make the number of hands or bust trying
during
each trip stake), and of course the game being played. Is that
correct?
Also, here I go again, but it seems to me that for rather small
numbers
of hands (say 2000, I pick 2000 because that is a good number for me
for a given daily trip), that the outcome distribution would be
skewed
toward the high end. So my question is when does it supposedly
become
normal as some on this forum seem to state? This distribution shape
keeps getting kicked around and has never been clear to me when it
becomes normal, if ever. I would think for small quantities (again
say 2000
hands) that it would never become normal. Jazbo seems to support
that it does not become normal.
Like I said, I have always had a problem with this, but (maybe it was
a poor explanation on the part of the posters) I was led to believe
that the outcome distribution would become normal as stated on vpFREE
by more than one poster on more than one occasion.
Somebody give me a good explanation please, when and where can you
assume normality? Or never?
A few questions:
Both of these will give me a graphical representation of the outcomes
of a given trip stake,using a given number of hands played during the
trip (I assume either you make the number of hands or bust trying
during each trip stake), and of course the game being played. Is
that correct?
On the money.
Also, here I go again, but it seems to me that for rather small
numbers of hands (say 2000, I pick 2000 because that is a good number
for me for a given daily trip), that the outcome distribution would
be skewed toward the high end. So my question is when does it
supposedly become normal as some on this forum seem to state? This
distribution shape keeps getting kicked around and has never been
clear to me when it becomes normal, if ever. I would think for small
quantities (again say 2000 hands) that it would never become normal.
Jazbo seems to support that it does not become normal.Like I said, I have always had a problem with this, but (maybe it was
a poor explanation on the part of the posters) I was led to believe
that the outcome distribution would become normal as stated on vpFREE
by more than one poster on more than one occasion.Somebody give me a good explanation please, when and where can you
assume normality? Or never?
As you observe (measured as "skewness"), in the short run we expect
far more losing sessions than winning sessions -- but the fewer
winning sessions will, at times, be large enough to balance things out
to where the game EV is preserved.
As you extend the number of hands played, cumulative results begin to
smooth out around the game EV; i.e. they start to distribute
themselves more equally around the mean. Once you look at play of
anywhere from 1MM to 10MM hands (the more volatile the game the more
hands needed), results become rather smoothly distributed around the
mean -- in loose terms, the point at which this is achieved is deemed
the "long term" of play; i.e. we reasonably expect that our cumulative
long term results will closely approximate game ER (in percentage terms).
However, even at this extreme point, play results do not meet the
criteria that define a "normal" distribution; they merely approximate
it. This may seem a bit paradoxical for we certainly expect that vp
probabilities overall adhere to a natural (aka "normal") distribution
-- there's nothing that distinguishes them in nature from simply
flipping coins.
But, what's at work here is that vp results don't measure a single
hit-or-miss outcome, but rather the aggregation of payoffs from a
discrete set of hands (Pair, 2 Pair ... RF). If you look at the
distribution of return derived from hitting pairs alone it quickly
takes on a normal distribution. The same is true for every other hand.
If each hand paid the same and occurred with the same frequency, then
the aggregate distribution of results would also take on a normal
distribution in short order. However, the disparate frequencies and
payouts cause each hand to unevenly contribute to the overall results
distribution ... and the hands that are least frequent and highest
paying are the ones that account for the bulk of a game's variance
statistic for, in addition to the high payout, they take the longest
for results to even out over the course of play.
Consequently, in the short run, when you aggregate the payout of
various hands you end up with a distribution that's misshapen -- where
a higher number of outcomes fall short of the mean and the results to
the right are "long tailed". It's only after the least frequently
occurring hands take on a normal distribution that the overall
distribution starts approximating "normality" itself.
A general rule of thumb is that for a game where there's only a single
low-frequency high-payout hand (e.g. Jacks vis-a-vis the RF), the long
term is achieved after 20-30 cycles of the hand. Where there are
several such hands (e.g. DDB, w/ RF & qA w/kicker), a greater number
of cycles are required.
- Harry
You need the Dunbar software for video poker analysis or Dancer's Video Poker for Winners.
....bl
\--- In vpFREE@yahoogroups.com, "Danton" <dabesq@...> wrote:
Hello,
I am considering some short term play above my normal limits to qualify
for some promotions, but I would like some help calculating the range
of my exposure to loss. For example, playing 9/6 JoB at $1, is there a
calculator that can tell me the range of possible outcomes for a given
number of hands? Say I play 200 hands, what does the distribution of
possible net results look like?Thanks!
You need the Dunbar software for video poker analysis or Dancer's Video Poker for Winners.
....bl
\--- In vpFREE@yahoogroups.com, "Danton" <dabesq@...> wrote:
Hello,
I am considering some short term play above my normal limits to qualify
for some promotions, but I would like some help calculating the range
of my exposure to loss. For example, playing 9/6 JoB at $1, is there a
calculator that can tell me the range of possible outcomes for a given
number of hands? Say I play 200 hands, what does the distribution of
possible net results look like?Thanks!
--- In vpFREE@yahoogroups.com, "Harry Porter" <harry.porter@...>
wrote:
> A few questions:
>
> Both of these will give me a graphical representation of the
outcomes
> of a given trip stake,using a given number of hands played during
the
> trip (I assume either you make the number of hands or bust
trying
> during each trip stake), and of course the game being played.
Is
> that correct?
On the money.
> Also, here I go again, but it seems to me that for rather small
> numbers of hands (say 2000, I pick 2000 because that is a good
number
> for me for a given daily trip), that the outcome distribution
would
> be skewed toward the high end. So my question is when does it
> supposedly become normal as some on this forum seem to state?
This
> distribution shape keeps getting kicked around and has never been
> clear to me when it becomes normal, if ever. I would think for
small
> quantities (again say 2000 hands) that it would never become
normal.
> Jazbo seems to support that it does not become normal.
>
> Like I said, I have always had a problem with this, but (maybe it
was
> a poor explanation on the part of the posters) I was led to
believe
> that the outcome distribution would become normal as stated on
vpFREE
> by more than one poster on more than one occasion.
>
> Somebody give me a good explanation please, when and where can
you
> assume normality? Or never?
As you observe (measured as "skewness"), in the short run we expect
far more losing sessions than winning sessions -- but the fewer
winning sessions will, at times, be large enough to balance things
out
to where the game EV is preserved.
As you extend the number of hands played, cumulative results begin
to
smooth out around the game EV; i.e. they start to distribute
themselves more equally around the mean. Once you look at play of
anywhere from 1MM to 10MM hands (the more volatile the game the more
hands needed), results become rather smoothly distributed around the
mean -- in loose terms, the point at which this is achieved is
deemed
the "long term" of play; i.e. we reasonably expect that our
cumulative
long term results will closely approximate game ER (in percentage
terms).
However, even at this extreme point, play results do not meet the
criteria that define a "normal" distribution; they merely
approximate
it. This may seem a bit paradoxical for we certainly expect that vp
probabilities overall adhere to a natural (aka "normal")
distribution
-- there's nothing that distinguishes them in nature from simply
flipping coins.But, what's at work here is that vp results don't measure a single
hit-or-miss outcome, but rather the aggregation of payoffs from a
discrete set of hands (Pair, 2 Pair ... RF). If you look at the
distribution of return derived from hitting pairs alone it quickly
takes on a normal distribution. The same is true for every other
hand.
If each hand paid the same and occurred with the same frequency,
then
the aggregate distribution of results would also take on a normal
distribution in short order. However, the disparate frequencies and
payouts cause each hand to unevenly contribute to the overall
results
distribution ... and the hands that are least frequent and highest
paying are the ones that account for the bulk of a game's variance
statistic for, in addition to the high payout, they take the longest
for results to even out over the course of play.Consequently, in the short run, when you aggregate the payout of
various hands you end up with a distribution that's misshapen --
where
a higher number of outcomes fall short of the mean and the results
to
the right are "long tailed". It's only after the least frequently
occurring hands take on a normal distribution that the overall
distribution starts approximating "normality" itself.A general rule of thumb is that for a game where there's only a
single
low-frequency high-payout hand (e.g. Jacks vis-a-vis the RF), the
long
term is achieved after 20-30 cycles of the hand. Where there are
several such hands (e.g. DDB, w/ RF & qA w/kicker), a greater number
of cycles are required.- Harry
Do Jazbo's curves agree with the above explanation? I may not have
asked the question properly, so I will try again,
Asuming I always play 1000 , 25c hands of NSUD (yes sometimes will
bust, but for now let that slide) For each of these sessions I take
$200 dollars with me and compute my win or loss as how much more than
$200 I take home that day, or how much less than $200 I take home
that day.
Then I make a distribution of my win, or loss for evey session/days
of 1000 hands. I would think that disrtibution would be centered
around a 0.3% of $1250 loss ($3.75 assuming perfect play) and skewed
toward the large win side and probably look like the curves that
Jazbo published. (use http://www.jazbo.com/ then vp
probabilities) It would further seem to me that the distribution
never would approximate the normal curve no matter if I did this 1000
hand session/trip for infinity.
Also, am I the only one who cannot understand this, or am I the only
one who will ask the question? Does anybody else care? Would the
group prefer that I take this discussion private?
deuceswild1000 wrote:
Do Jazbo's curves agree with the above explanation? I may not have
asked the question properly, so I will try again,Asuming I always play 1000 , 25c hands of NSUD (yes sometimes will
bust, but for now let that slide) For each of these sessions I take
$200 dollars with me and compute my win or loss as how much more than
$200 I take home that day, or how much less than $200 I take home
that day.Then I make a distribution of my win, or loss for evey session/days
of 1000 hands. I would think that disrtibution would be centered
around a 0.3% of $1250 loss ($3.75 assuming perfect play) and skewed
toward the large win side and probably look like the curves that
Jazbo published. (use http://www.jazbo.com/ then vp
probabilities) It would further seem to me that the distribution
never would approximate the normal curve no matter if I did this 1000
hand session/trip for infinity.
You would never expect that vp results for limited hand sessions/trips
would have approximate a normal curve. Over the short term, sessions
are sharply defined by whether or not you hit infrequent, large paying
hands.
At the extreme, you'll notice in Jazbo's charts that over a reasonably
large number of hands (yet, still in the "short run") the curve begins
to show two peaks -- corresponding to those sessions in which royals
were hit and those in which they weren't.
This disparate influence of hands that occur frequently yet are low
paying vs. those that are infrequent yet are high paying serves to
cause the skewness that is observed in short run sessions. It's only
when you chart individual runs of over 1+ mil hands that the impact of
this difference smooths out and the charts begin to approximate a
normal distribution.
- Harry
--- In vpFREE@yahoogroups.com, "Harry Porter" <harry.porter@...>
wrote:
deuceswild1000 wrote:
> Do Jazbo's curves agree with the above explanation? I may not
have
> asked the question properly, so I will try again,
>
> Asuming I always play 1000 , 25c hands of NSUD (yes sometimes
will
> bust, but for now let that slide) For each of these sessions I
take
> $200 dollars with me and compute my win or loss as how much more
than
> $200 I take home that day, or how much less than $200 I take
home
> that day.
>
> Then I make a distribution of my win, or loss for evey
session/days
> of 1000 hands. I would think that disrtibution would be
centered
> around a 0.3% of $1250 loss ($3.75 assuming perfect play) and
skewed
> toward the large win side and probably look like the curves that
> Jazbo published. (use http://www.jazbo.com/ then vp
> probabilities) It would further seem to me that the
distribution
> never would approximate the normal curve no matter if I did this
1000
> hand session/trip for infinity.
You would never expect that vp results for limited hand
sessions/trips
would have approximate a normal curve. Over the short term,
sessions
are sharply defined by whether or not you hit infrequent, large
paying
hands.
At the extreme, you'll notice in Jazbo's charts that over a
reasonably
large number of hands (yet, still in the "short run") the curve
begins
to show two peaks -- corresponding to those sessions in which
royals
were hit and those in which they weren't.
This disparate influence of hands that occur frequently yet are low
paying vs. those that are infrequent yet are high paying serves to
cause the skewness that is observed in short run sessions. It's
only
when you chart individual runs of over 1+ mil hands that the
impact of
this difference smooths out and the charts begin to approximate a
normal distribution.- Harry
That is what I always thought. I must have mis-stated my question
in my original posting.
--- In vpFREE@yahoogroups.com, "deuceswild1000" <deuceswild1000@...>
wrote:
--- In vpFREE@yahoogroups.com, "Harry Porter" <harry.porter@>
wrote:
>
> > A few questions:
> >
> > Both of these will give me a graphical representation of the
outcomes
> > of a given trip stake,using a given number of hands played
during
the
> > trip (I assume either you make the number of hands or bust
trying
> > during each trip stake), and of course the game being played.
Is
> > that correct?
>
> On the money.
>
>
> > Also, here I go again, but it seems to me that for rather small
> > numbers of hands (say 2000, I pick 2000 because that is a good
number
> > for me for a given daily trip), that the outcome distribution
would
> > be skewed toward the high end. So my question is when does it
> > supposedly become normal as some on this forum seem to state?
This
> > distribution shape keeps getting kicked around and has never
been
> > clear to me when it becomes normal, if ever. I would think for
small
> > quantities (again say 2000 hands) that it would never become
normal.
> > Jazbo seems to support that it does not become normal.
> >
> > Like I said, I have always had a problem with this, but (maybe
it
was
> > a poor explanation on the part of the posters) I was led to
believe
> > that the outcome distribution would become normal as stated on
vpFREE
> > by more than one poster on more than one occasion.
> >
> > Somebody give me a good explanation please, when and where can
you
> > assume normality? Or never?
>
>
> As you observe (measured as "skewness"), in the short run we
expect
> far more losing sessions than winning sessions -- but the fewer
> winning sessions will, at times, be large enough to balance
things
out
> to where the game EV is preserved.
>
> As you extend the number of hands played, cumulative results
begin
to
> smooth out around the game EV; i.e. they start to distribute
> themselves more equally around the mean. Once you look at play of
> anywhere from 1MM to 10MM hands (the more volatile the game the
more
> hands needed), results become rather smoothly distributed around
the
> mean -- in loose terms, the point at which this is achieved is
deemed
> the "long term" of play; i.e. we reasonably expect that our
cumulative
> long term results will closely approximate game ER (in percentage
terms).
>
> However, even at this extreme point, play results do not meet the
> criteria that define a "normal" distribution; they merely
approximate
> it. This may seem a bit paradoxical for we certainly expect that
vp
> probabilities overall adhere to a natural (aka "normal")
distribution
> -- there's nothing that distinguishes them in nature from simply
> flipping coins.
>
> But, what's at work here is that vp results don't measure a single
> hit-or-miss outcome, but rather the aggregation of payoffs from a
> discrete set of hands (Pair, 2 Pair ... RF). If you look at the
> distribution of return derived from hitting pairs alone it quickly
> takes on a normal distribution. The same is true for every other
hand.
>
> If each hand paid the same and occurred with the same frequency,
then
> the aggregate distribution of results would also take on a normal
> distribution in short order. However, the disparate frequencies
and
> payouts cause each hand to unevenly contribute to the overall
results
> distribution ... and the hands that are least frequent and highest
> paying are the ones that account for the bulk of a game's variance
> statistic for, in addition to the high payout, they take the
longest
> for results to even out over the course of play.
>
> Consequently, in the short run, when you aggregate the payout of
> various hands you end up with a distribution that's misshapen --
where
> a higher number of outcomes fall short of the mean and the
results
to
> the right are "long tailed". It's only after the least frequently
> occurring hands take on a normal distribution that the overall
> distribution starts approximating "normality" itself.
>
> A general rule of thumb is that for a game where there's only a
single
> low-frequency high-payout hand (e.g. Jacks vis-a-vis the RF), the
long
> term is achieved after 20-30 cycles of the hand. Where there are
> several such hands (e.g. DDB, w/ RF & qA w/kicker), a greater
number
> of cycles are required.
>
> - Harry
>Do Jazbo's curves agree with the above explanation? I may not have
asked the question properly, so I will try again,Asuming I always play 1000 , 25c hands of NSUD (yes sometimes will
bust, but for now let that slide) For each of these sessions I
take
$200 dollars with me and compute my win or loss as how much more
than
$200 I take home that day, or how much less than $200 I take home
that day.Then I make a distribution of my win, or loss for evey session/days
of 1000 hands. I would think that disrtibution would be centered
around a 0.3% of $1250 loss ($3.75 assuming perfect play) and
skewed
toward the large win side and probably look like the curves that
Jazbo published. (use http://www.jazbo.com/ then vp
probabilities) It would further seem to me that the distribution
never would approximate the normal curve no matter if I did this
1000
hand session/trip for infinity.
Also, am I the only one who cannot understand this, or am I the
only
one who will ask the question? Does anybody else care? Would the
group prefer that I take this discussion private?
I took a look at the Jazbo curves you referenced and they look as I
would expect. If you agree with them I guess I don't understand what
it is you don't understand.
The more hands you play, the more the curve looks like a normal
distribution (long run) and the fewer hands you play, the less the
curve looks normal because the random streakiness overpowers the law
of large numbers which isn't large enough yet. This is the basis of
the disagreement between the the optimal play forever folks (vast
majority) versus the stop while you're ahead, sometime go for 4oaks,
put floors under your positive gains, etc. (black swan tiny
minority). Under various titles this has and will continue to be
covered many times on this and all other poker/video poker blogs.
I will say the fact that you are asking this question means you are
trying to dig deeper than just learning optimal play rules (which is
the mandatory starting point) and in my opinion this will over time
make you a more successful player than you would have otherwise
been. Denny
I took a look at the Jazbo curves you referenced and they look as I
would expect. If you agree with them I guess I don't understand
what
it is you don't understand.
The more hands you play, the more the curve looks like a normal
distribution (long run) and the fewer hands you play, the less the
curve looks normal
(Again, I am probably wrong or stating this wrong, but here goes.)
If I play 1,000,000 hands at perfect play, I would expect my final
return (be it positive or negative based on the game I am playing)
to be close to the expected return of that game.
However, after a 1,000,000 hand session, I only have one data point,
which no way constitutes a distribution or a curve, which ever you
prefer to call it.
So to expand on that:
If I had quite a few (say 50) of these 1,000,000 hand session data
points, then I would expect that they would be approach being
normally distributed and the variance of the sample would be very
small.
Similarly if I only do 1000 hand sessions multiple times, Iwould get
a multiple of data points that I can put into a distribution and would
eventually expect the average of those sessions data points to
approach the expected return. The variance of this distribution
would be much larger than the one for 1,000,000 session
distribution. Also as documented by Jazbo, the distribution would
not be normally distributed
I may be playing on words here, but I think the association of large
number of hands and the assumption of a normal distribution leaves
out what I just covered above.
Why do I bring it up, because, I think that unless one has the
further explanation, they can make a wrong conclusion. I am hoping
that someone can benefit from my own intial confusion of the
association of large number of hands and normal distibution as is
often state on these forums.
I am open to any and all agreements or disagreements as a learning
experience.
--- In vpFREE@yahoogroups.com, "denflo60" <dennis.florence@...> wrote:
Calculation of risk of loss I am new to this vp gambling why would you care
about risk of loss you x amount of money you win or lose you can count whay
you won or loss what is the purpose of calculating the win loss?
_____
From: vpFREE@yahoogroups.com [mailto:vpF…@…com] On Behalf Of
Harry Porter
Sent: Friday, May 18, 2007 6:42 AM
To: vpFREE@yahoogroups.com
Subject: [vpFREE] Re: Calculation of Risk of Loss
deuceswild1000 wrote:
Do Jazbo's curves agree with the above explanation? I may not have
asked the question properly, so I will try again,Asuming I always play 1000 , 25c hands of NSUD (yes sometimes will
bust, but for now let that slide) For each of these sessions I take
$200 dollars with me and compute my win or loss as how much more than
$200 I take home that day, or how much less than $200 I take home
that day.Then I make a distribution of my win, or loss for evey session/days
of 1000 hands. I would think that disrtibution would be centered
around a 0.3% of $1250 loss ($3.75 assuming perfect play) and skewed
toward the large win side and probably look like the curves that
Jazbo published. (use http://www.jazbo.com/> com/ then
vp
probabilities) It would further seem to me that the distribution
never would approximate the normal curve no matter if I did this 1000
hand session/trip for infinity.
You would never expect that vp results for limited hand sessions/trips
would have approximate a normal curve. Over the short term, sessions
are sharply defined by whether or not you hit infrequent, large paying
hands.
At the extreme, you'll notice in Jazbo's charts that over a reasonably
large number of hands (yet, still in the "short run") the curve begins
to show two peaks -- corresponding to those sessions in which royals
were hit and those in which they weren't.
This disparate influence of hands that occur frequently yet are low
paying vs. those that are infrequent yet are high paying serves to
cause the skewness that is observed in short run sessions. It's only
when you chart individual runs of over 1+ mil hands that the impact of
this difference smooths out and the charts begin to approximate a
normal distribution.
- Harry
[Non-text portions of this message have been removed]
john wrote:
Calculation of risk of loss I am new to this vp gambling why would you
care
about risk of loss you x amount of money you win or lose you can count
why
you won or loss what is the purpose of calculating the win loss?
Most of us approach play by finding an enjoyable and comfortable game
and denomination and hit the casino without terribly obsessing over what
we might lose during the visit. We simply come with a given amount of
cash and call it a day should we lose, or simply tire of the play. If
we end with a win, so much the better.
However, there are occasions where a player has good cause to be more
concerned with their loss risk. An example is when a player considers
moving up in denomination temporarily. The original poster in this
thread posed a scenario in which he was considering playing at a higher
denomination for a while in order to qualify for stronger promotions.
There's a hazard in such a move.
While playing at your typical play level (say $.25), you experience
swings that over time you expect to largely even out. However, if you
play at an increased denomination for awhile and suffer a sizable loss,
then move back to your standard denom, it's now going to take some
rather strong luck to recoup the loss suffered at the higher denom. In
fact, there's good reason to look on the situation and see any short
term higher denom loss as a PERMANENT loss, and not just a temporary
swing.
So, under such circumstances, it's prudent to have a decent quantitative
sense of the loss risk you face before starting that higher denomination
play -- reducing the chance that you boldly step into the deeper waters
and suddenly find the tide is sweeping you away.
- Harry
(btw, it's possible I didn't properly read what you were asking -- if
I'm off the mark, feel free to set me straight and give me a second
shot.)
I've never been to AC but am thinking of going on a HET offer.
Which of the 4 HET properties do you recommend? Are they within walking
distance from each other?
Thanks
JAS
Ceasars & Ballys are next to each other mid boardwalk. Showboat is
about 5 blocks from them. Harrahs is in the marina. I prefer Harrahs
first, Showboat, Ballys & then Ceasars.--- In
vpFREE@yahoogroups.com, "snow" <jsfs@...> wrote:
I've never been to AC but am thinking of going on a HET offer.
Which of the 4 HET properties do you recommend? Are they within
walking
distance from each other?
Thanks
JAS
--- In vpFREE@yahoogroups.com, "deuceswild1000" <deuceswild1000@...>
wrote:
> I took a look at the Jazbo curves you referenced and they look as
I
> would expect. If you agree with them I guess I don't understand
what
> it is you don't understand.
>
> The more hands you play, the more the curve looks like a normal
> distribution (long run) and the fewer hands you play, the less
the
> curve looks normal
(Again, I am probably wrong or stating this wrong, but here goes.)
If I play 1,000,000 hands at perfect play, I would expect my final
return (be it positive or negative based on the game I am playing)
to be close to the expected return of that game.However, after a 1,000,000 hand session, I only have one data
point,
which no way constitutes a distribution or a curve, which ever you
prefer to call it.So to expand on that:
If I had quite a few (say 50) of these 1,000,000 hand session data
points, then I would expect that they would be approach being
normally distributed and the variance of the sample would be very
small.Similarly if I only do 1000 hand sessions multiple times, Iwould
get
a multiple of data points that I can put into a distribution and
would
eventually expect the average of those sessions data points to
approach the expected return. The variance of this distribution
would be much larger than the one for 1,000,000 session
distribution. Also as documented by Jazbo, the distribution would
not be normally distributed
deuceswild1000: Now I see what you're saying and it has given me
reason to pause and reflect. I must have missed your early posts and
therefore the crux of your question. This old engineer has to go and
find an old stat book and look this up. Maybe I'll buy "Statistics
for Dummies". It is a real book, and right now the way you've
phrased your question I am hesitant to respond instantaneously and I
feel like a dummy. I love these kind of questions. Off to Vegas
next week and maybe a visit to the "Gambler's Bookstore" is in
order. Denny
I may be playing on words here, but I think the association of
large
--- In vpFREE@yahoogroups.com, "denflo60" <dennis.florence@> wrote:
number of hands and the assumption of a normal distribution leaves
out what I just covered above.Why do I bring it up, because, I think that unless one has the
further explanation, they can make a wrong conclusion. I am hoping
that someone can benefit from my own intial confusion of the
association of large number of hands and normal distibution as is
often state on these forums.I am open to any and all agreements or disagreements as a learning
experience.
--- In vpFREE@yahoogroups.com, "deuceswild1000" <deuceswild1000@...>
wrote:
> I took a look at the Jazbo curves you referenced and they look as
I
> would expect. If you agree with them I guess I don't understand
what
> it is you don't understand.
>
> The more hands you play, the more the curve looks like a normal
> distribution (long run) and the fewer hands you play, the less
the
> curve looks normal
(Again, I am probably wrong or stating this wrong, but here goes.)
If I play 1,000,000 hands at perfect play, I would expect my final
return (be it positive or negative based on the game I am playing)
to be close to the expected return of that game.However, after a 1,000,000 hand session, I only have one data
point,
which no way constitutes a distribution or a curve, which ever you
prefer to call it.So to expand on that:
If I had quite a few (say 50) of these 1,000,000 hand session data
points, then I would expect that they would be approach being
normally distributed and the variance of the sample would be very
small.Similarly if I only do 1000 hand sessions multiple times, Iwould
get
a multiple of data points that I can put into a distribution and
would
eventually expect the average of those sessions data points to
approach the expected return. The variance of this distribution
would be much larger than the one for 1,000,000 session
distribution. Also as documented by Jazbo, the distribution would
not be normally distributedI may be playing on words here, but I think the association of
large
number of hands and the assumption of a normal distribution leaves
out what I just covered above.Why do I bring it up, because, I think that unless one has the
further explanation, they can make a wrong conclusion. I am hoping
that someone can benefit from my own intial confusion of the
association of large number of hands and normal distibution as is
often state on these forums.I am open to any and all agreements or disagreements as a learning
experience.
deuceswild1000: I think my own understanding gets muddled when I
think of precise numbers. So I simply try to qualitatively look at
Jazbo type curves as short term <1000 hands and long term >10M
hands. For low number of hands any specific session will generate a
curve that is less close a to being normal while long term sessions
generate curves that are likely very close to normal. More specific
than that I will not get. I still feel that this is what causes the
interesting short term vs. long term differences of opinion between
the optimal play traditionalists and the short term folks who feel
they can do better by many varied approaches. As I recall in cases
where things are not normal curves, except in very specific instances
(poisson for example), variance discussions are really not
applicable. Denny
--- In vpFREE@yahoogroups.com, "denflo60" <dennis.florence@> wrote:
deuceswild1000: I think my own understanding gets muddled when I
think of precise numbers. So I simply try to qualitatively look at
Jazbo type curves as short term <1000 hands and long term >10M
hands. For low number of hands any specific session will generate
a
curve that is less close a to being normal while long term sessions
generate curves that are likely very close to normal.
My point is I do not see how a session can generate a distribution.
A session to me would generate a single data to be put into a
distibution
of sessions of a given hand size.
This distribution would then be able to be analyzed for mean, mode,
variance, skewness, shape, and etc.
So, that is why I keep asking: How can A large number of hand
session approach normal.
So far any explanation I have received it not able to break it down
to my level of understanding.
Another way of asking my question is if A large session generates a
distribution that approaches normal, then using a histogram for
illustration purposes, what goes on the x axis. I assume frequency
goes on the y axis.
--- In vpFREE@yahoogroups.com, "denflo60" <dennis.florence@...> wrote:
--- In vpFREE@yahoogroups.com, "deuceswild1000" <deuceswild1000@...>
wrote:
> deuceswild1000: I think my own understanding gets muddled when I
> think of precise numbers. So I simply try to qualitatively look
at
> Jazbo type curves as short term <1000 hands and long term >10M
> hands. For low number of hands any specific session will
generate
a
> curve that is less close a to being normal while long term
sessions
> generate curves that are likely very close to normal.
My point is I do not see how a session can generate a distribution.
A session to me would generate a single data to be put into a
distibution
of sessions of a given hand size.This distribution would then be able to be analyzed for mean, mode,
variance, skewness, shape, and etc.So, that is why I keep asking: How can A large number of hand
session approach normal.So far any explanation I have received it not able to break it down
to my level of understanding.Another way of asking my question is if A large session generates a
distribution that approaches normal, then using a histogram for
illustration purposes, what goes on the x axis. I assume frequency
goes on the y axis.
deuceswild1000: There are exellent chapters on the subject of normal
(and not normal) distributions in the recent books by Taleb and
Mandelbrot. Also wonderful discussions how very smart people have
misused normal distributions. There are also is a discussion how you
can generate normal Or not so normal curves from data points which I
have implied to mean hands in a session as well as entire sessions.
You might want to look into these books which are available at
Amazon. Also take a peek at the archives from "2+2 forums" which you
can reach through Google and even www.videopoker/forums.com where
discussions get a bit wild and wooly. Also try googling around words
such as gambling and probability, long and short run pobability,
cycles in randomness etc. etc. etc.
I have struggled mightily to better understand this whole subject of
what is and where can normal ditributions be used and what is most
important over the long and short run. Despite what the vast
majority of people say, there is still much to be learned here in my
opinion. The very smartest most learned people are unable to prove
that short term probablities cannot be exploited. They may be
correct when they say it, but they cannot prove it mathematically.
At least I have found nothing yet in the literature that proves it
and no one can give me a reference that proves it.
The study you are attempting to do is a great thing. I would expect
an answer lies in the bowels of IGT Corp. or associated companies but
as I have said before, it is locked in the same safe as the Coca Cola
formula for the same reason the Coke formula is.
I'm all done now with this subject now until someone else in one of
these forums raises it up again in a slightly different context.
Denny
--- In vpFREE@yahoogroups.com, "denflo60" <dennis.florence@> wrote: