vpFREE2 Forums

Test of Randomness, ER, etc.

--- In vpFREE@yahoogroups.com, "jaydavidson118" <jaydavidson118@y...>
wrote:

One important characteristic that effects the

ability to apply CLT is how close the underlying distribution is to
the Normal distribution. The answer in our case is not very close.
The normal distribution is a symetric distribution.

::::::::::::::::
I am somewhat familiar with the CLT, but no way an expert. But as I
understand it the power of the CLT is it does apply to non symetrical
distributions, even very non symetrical distributions. But the
sample sizes need to be large. There is a subtle aspect to this I
havent seen discussed, where VP is concerned. We talk about a
session of 1 million games, and that seems like a very large sample.
But in JoB there are only 9 possible outcomes, a loss, or 8 kinds of
wins. That is a small number. So is that the number that really
counts when deciding whether the CLT applies, or is it the number of
games played? If it is the former, then it seems to mean the CLT
does not really apply to VP, in a strict mathematical sense.
Personally, I do think it does apply to VP because I have run dozens
of large session tests, from 50,000 games up to 1 million, that seem
to confirm it does apply to VP.

--- In vpFREE@yahoogroups.com, "jaydavidson118" <jaydavidson118@y...>
wrote:

It takes a little knowledge to put numbers into a statistical

package

and print them out, it takes considerably more knowledge to know

how

to intrepret the results, and to know whether they are meaningful.

I

don't see much indication of the latter. Is that too forceful?

from your comments it's obvious you don't read the entire message
that might be a good place to start
just a suggestion

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

However, it occurs to me that there may be a problem with suggesting
that given an adverse outcome (the $340 loss, in your example) that
has less than a 2% likelihood, you'd have less than a 2% confidence

in

the machine's fairness.

After all, throw just 60 people at a fair machine for a trial of

2000

hands each and it's expected that on average one of the people will
cry "the machine is fixed!" using the criteria you suggest (mind

you,

I realize I'm taking a little liberty with your actual wording).

well, that's what 2% means
if you want 1% or some other arbitrary number, you can certainly
modify the procedure to your liking

You've given Chris a decent practical test of fairness. But it's
hardly a sufficient one to take to the Casino Control Commission as
the basis of a complaint.

I disagree, I think it's a perfectly legitimate complaint, let the
board investigate, perhaps other people have had similar results ...

Plus, in his shoes I'd find it difficult to
translate it directly into some type of acid test (what to

play/where

to play), given the inherent limitations.

actually, it's a perfectly good acid test, if your results are that
bad, time to find a new machine (true there is a 2% chance the machine
is actually ok, but why waste time? ...)

That said, if he or anyone else wishes to maintain the necessary
detailed records, your methodology permits a specific measurement of
how likely a given adverse result is and, in comparing results from
one casino to the next, may be a meaningful guide that suggests one

or

more casinos that it wouldn't hurt to avoid.

the necessary detailed records are pretty spartan, just record the
casino and machine number and keep track of the first 2000 hands
(often the card club reader tells you that or if you want there are
indicators under the machine that indicate it) and keep track of your
coin in and coin out, all you need to know is your running loss

Harry Porter wrote:

> However, it occurs to me that there may be a problem with
> suggesting that given an adverse outcome (the $340 loss, in your
> example) that has less than a 2% likelihood, you'd have less than
> a 2% confidence in the machine's fairness.

iggy replied:

well, that's what 2% means
if you want 1% or some other arbitrary number, you can certainly
modify the procedure to your liking

Iggy, I think the best case here is that we're going to agree to
disagree with each other -- strongly.

Extending your logic -- I might ask, once the determination is made
that there's over a 98% likelihood that the machine is biased, what's
the most likely amount of that bias? Your interpretation would lead
me to decide that it's most likely that the machine is negatively
biased by 8% to 9%, based upon this single trial, since that's the
bias that would most likely produce this session result.

Correct me, if you see it otherwise.

I then see you suggesting that I should reasonably proceed to the
casino commission with my evaluation and tell them that the machine is
likely gaffed, most probably by more than 8%. (Understand that I have
a problem saying the machine is gaffed if I can't quantify it more
strongly than "by at least a fraction of a percent, and probably more".)

If I'm on track in what you're saying by implication, I won't deem it
"silly", as jay did. But I will say that it's not reasonable.

If it should be the case that I'm misreading/distorting what you've
said, I'm not sure it's productive the explore this further, But
until I understand otherwise, I'm not able to accept your extension of
the probability of a given result to a precise statement of confidence
in machine fairness/unfairness.

- Harry

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

Harry Porter wrote:
> > However, it occurs to me that there may be a problem with
> > suggesting that given an adverse outcome (the $340 loss, in

your

> > example) that has less than a 2% likelihood, you'd have less

than

> > a 2% confidence in the machine's fairness.

iggy replied:
> well, that's what 2% means
> if you want 1% or some other arbitrary number, you can certainly
> modify the procedure to your liking

Iggy, I think the best case here is that we're going to agree to
disagree with each other -- strongly.

Extending your logic -- I might ask, once the determination is made
that there's over a 98% likelihood that the machine is biased,

what's

the most likely amount of that bias? Your interpretation would lead
me to decide that it's most likely that the machine is negatively
biased by 8% to 9%, based upon this single trial, since that's the
bias that would most likely produce this session result.

now you're asking for more conclusions, not only do you want to
determine if a machine is biased, you want to know by what amount?
that's a much more complex issue, you have to know the variance of
your 2000 hand result as well as the average return ... then you've
got what in statistics they call sampling ...

Correct me, if you see it otherwise.

I then see you suggesting that I should reasonably proceed to the
casino commission with my evaluation and tell them that the machine

is

likely gaffed, most probably by more than 8%. (Understand that I

have

a problem saying the machine is gaffed if I can't quantify it more
strongly than "by at least a fraction of a percent, and probably

more".)

just tell them you suspect it is gaffed, let them determine if it
really is and by what amount
that's their job, if you insist you can do further studies, but why?
is it worth it?
once you suspect a machine is even 1% less than it should be, is it
really worth determining how bad it is? personally i would just report
it and move on
in my book a gaffed machine isn't playable, i don't really care by how
much it's gaffed
(excluding a few rare cases where a consistent gaff was beatable with
strategy modification)
it's the same as blackjack, if i know a dealer is dealing seconds, i
don't really care how often they are doing it and on what hands
likewise, legally, a gaffed machine is a gaffed machine, the amount of
gaffing has no legal results

If I'm on track in what you're saying by implication, I won't deem

it

"silly", as jay did. But I will say that it's not reasonable.

If it should be the case that I'm misreading/distorting what you've
said, I'm not sure it's productive the explore this further, But
until I understand otherwise, I'm not able to accept your extension

of

the probability of a given result to a precise statement of

confidence

in machine fairness/unfairness.

- Harry

on the topic of gaffs, in "All the Best of Video Poker Times" by Dan
Paymar, there is a report of fixed blackjack machines at a Rhode
Island racetrack, the player noticed the dealer got blackjack with an
ace more than random results would allow, the player wisely altered
strategy and always took insurance and beat the machines badly

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

Extending your logic -- I might ask, once the determination is made
that there's over a 98% likelihood that the machine is biased,

what's

the most likely amount of that bias?

i think i have a quick first cut at this:
if you want 68% confidence (1sd) of measuring the actual average
return +/-10%, you would need a sample size of 100N0 hands

one more point, if you keep it simple you can take a complaint like
this to your host
just don't tell them anything that would make them suspect you know
what you are talking about
just tell them your loses and the machine number and tell them you
think it's a fixed machine and you're thinking about going to the
regulation board about it, basically you're mad as hell and you're not
gonna take it anymore
don't lie about your machine results, they can most likely look it up
on computer
if they ask you why you think it's fixed just tell them you've never
lost that much before, ask them, doesn't it seem like an unusually
large loss? but if they ask you if you've ever read this group or
heard of dancer, paymar, scott or wang you better act dumb
you might get some compensation
stuff like this happens all the time in the pit

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

one more point, if you keep it simple you can take a complaint like
this to your host
just don't tell them anything that would make them suspect you know
what you are talking about
just tell them your loses and the machine number and tell them you
think it's a fixed machine and you're thinking about going to the
regulation board about it, basically you're mad as hell and you're

not

gonna take it anymore
don't lie about your machine results, they can most likely look it

up

on computer
if they ask you why you think it's fixed just tell them you've

never

lost that much before, ask them, doesn't it seem like an unusually
large loss? but if they ask you if you've ever read this group or
heard of dancer, paymar, scott or wang you better act dumb
you might get some compensation
stuff like this happens all the time in the pit

Thanks Iguana,

Good idea! If you are a regular player and have a really bad
session they just might want to ease the pain a little.

I used your variance numbers for AA and found to my surprise the
variance is almost twice that of FPDW (6.49 versus 3.67). Even
though the ER for AA is 96% versus 95% for FPDW, the lower limit for
quarters is $385 versus $340. In my limited experience of playing
AA here in Kansas City the downward swings in a short session were
surprising (upward swings are no problem of course), but I was
comparing it to my 3 trips to LV where I played FPDW.

Let's go back to the question of rigging poker machines. There is
another possibility that has not been discussed. I am sure the
chips and software are tested, but non fatal errors could happen
that reduced ER. The machine makers and the casinos would probably
keep a tight lid on such matters, but Murphy's Law guarantees that
mistakes happen. The casinos monitor the hold on machines and would
probably jump on any machine that was paying too much, but how
quickly would they check on a machine that was paying too little?

Comments from someone who has knowledge of casino procedures would
be appreciated.

Chris