What Would It Take???

OK I agree. I also think dealt hands is a good place to start. We need to eliminate the effects of personal strategy. I see several ways to do this.

1. Go off dealt paying hands as a first step.

2. Also record 1 card draws like 4FL, 2-Pair, 4 to the straight etc...

3. Then compare made five-card hands like Full House that one gets drawing to 2-Pair. (Not all your Full Houses, only the ones you get drawing 1)

In this manner it wouldn't matter how people played or what strategy they used. Do you agree?

Thank you Marcus. If I have the time I will check out that book. It looks like it's got what we need.

Each "card" has a unique name, because the suits can be represented as (s,d,c,h). The "card" record can be keyed into Excel, to assign a number to each card, e.g. Excel will transform "Ks" into the number 13.

Because there are 52 possible "cards," the number of observations will need to be quite large to get a valid result. At least 200 dealt hands and subsequent draws, I would think.

I'm making some good progress on the other forums. I'll give everyone a heads up on major developments when they occur.

On vpFREE the suggestion came up that a video camera be used to record the hands and that one could track all five dealt cards and all drawn cards for simple 1 in 52 randomness.

Opinions?

Is anyone sure it's legal given todays rules on cameras in casinos. I know they changed the rules recently, but I'm still living in the past.

Out for rest of day.

I have no idea if what I'm attempting is even possible or what the end result will be, and like most good science, "research is what I'm doing when I have no idea what I'm doing."

Thank you all for your ideas, keep em-coming and I'll post a "what's left to figure out" in the morning.

~FK

It's simple, Frank. Record only the cards you discard and the cards they get replaced with:

7s -> 2c
4h -> Qh

etc., etc.

Then test the sequence 7s,2c,4h,Qh.... for non-randomness. You can break it into two sequences, 7,2,4,12... and s,c,h,h... You won't need a huge number of observations to determine whether the suits are non-random, 100 cards should be enough. But for the ranks you might need 1,000.

Designing a non-random deal that would pass this test would be child's play. E.g. without replacement, deal out y cards, starting at card 53-y, using non-random method X. Increment y with a random number, return and repeat.

Major positive development.

I just found a friend that is a visual basic programming expert with skills far beyond anything I've ever seen, that will work for food. OK gourmet comps, but hay it's still food. This will greatly improve the usability and functionality of the final product.

Once we lock down what we want coded, he can do it in the blink of an eye. I'm meeting him tonight.

~FK

The problem is what MrMarcus alluded to:

If the cards are non-random it is most likely fraud. In other words, on purpose. If the chip was criminally set to alter the number of royals with 4 to a royal from 1/47 to 1/70 it would be tough for a program to detect that. Could the program detect that? If not, the exercise is of limited utility.

Serial Correlation Coefficient

This quantity measures the extent to which each value in the sequence depends upon the previous value. For random sequences, this metric (which can be positive or negative) will be close to zero. A non-random sequence such as a text file will yield a serial correlation coefficient of about 0.5. Predictable data will exhibit serial correlation coefficients approaching 1.


Autocorrelation

The correlation between values of the same variable at different times.
Sometimes referred to as serial correlation.
Autocorrelation coefficient is calculated by substituting lagged data pairs into the formula for the Pearson product-moment correlation coefficient.
Autocorrelation function is the collection of autocorrelation coefficients computed for various lags.
Function always begins with an autocorrelation coefficient of 1, since a series of unshifted data will exhibit perfect correlation with itself.
Function will decay towards zero as lag increases.
Used to detect non-randomness in data.