Thoughts For Beginners

[FloridaPhil]

[quote=Carcounter]If you play a 99.54%game like 9/6 JOB your long term results will
always be dependent on whether you hit an average,below average, or
higher than average number of royals[/quote]It takes the proper amount of all hands to come out even with the ER.  Personally, I believe more return is lost due to human errors than anything that happens with the RNG.  As much as we don't want to admit it, it's much harder to play computer prefect for months or years on end than is commonly believed.

[New2vp]

1 out of 20 hands----"can fail"---5% this, 5% that---New2vp....your circular statistical logic only proves more auto-response sophistication of how paytables/returns of vp games can be technically/legally manipulated.....With regards to you gobbledy-gooking an auto-defense of a commercial institution called Bob Dancer--undermines your credibility further....But, thanks for your unintended support for my legitimate concern with your circular input. Every time Dancer is put on the spot or negatively criticized, he historically becomes math-wonkishly belittling as well....Rapidbison, sorry if you misunderstood what I wrote.  I thought you were confused before.  Sorry for trying to improve your information set.  I now understand that it is more important for you to hold onto your preconceived notions.  No problem for me either way.  I did not learn math or statistics from Mr. Dancer, nor would I support anything he says if I wasn't able to understand things on my own terms.  So, I wouldn't expect anyone else to do that either.  Similarly I wouldn't believe anything simply because you wrote it.  Ironically, in the post that you are criticizing for defending Bob Dancer, I was writing about something that I don't recall him ever talking about, though I suppose he could have written about it and I just missed it.I also didn't make up the regulations that I referenced for you that require chi-squared goodness-of-fit tests.  I do realize that most may have never heard of such a test, let alone understood how to conduct one.  I wish you good health and hope you never are stricken with a disease that you believe cannot exist simply because you have never heard of it or fail to understand how it is transmitted.  I would still recommend that you seek out someone with more training in the field than you have had.There was nothing intended to belittle anyone in the previous post, mathematically or otherwise.  Though there are no direct insults in this one, you may appropriately judge this response to be more sarcastic,.  I did keep math out of it.  Certainly it is still not as mean-spirited as your response to my earlier explanation.If you want to believe that 1 out of 20 hands are not dealt randomly, I'm not certain how that helps you (unless maybe it explains why your own results have not been as positive as you would have hoped), but you are welcome to that belief.  I imagine that there are many Class II machines that are even worse than that.  If you end up with a hypothesis of how those hands are dealt on Class III machines, if not randomly, please be certain to post it to educate the rest of it.  I know now that you are uninterested in responses to your posts in areas that explain phenomena that you do not understand. However, if you truly become interested in other points of view, please let me know.  I don't read every post the way I used to do, so I might miss it.  Please be patient and have a great rest of the weekend! 

[asteroid]

I guess they use the Chi-Square test because Fisher's Exact test is still computationally too intensive and there are enough data points to converge on the results of a an exact test (give it 5 years and moore's law will set things right).
[QUOTE=rapidbison]1 out of 20 hands----"can fail"---5% this, 5% that---New2vp....your circular statistical logic only proves more auto-response sophistication of how paytables/returns of vp games can be technically/legally manipulated.....With regards to you gobbledy-gooking an auto-defense of a commercial institution called Bob Dancer--undermines your credibility further....But, thanks for your unintended support for my legitimate concern with your circular input. Every time Dancer is put on the spot or negatively criticized, he historically becomes math-wonkishly belittling as well....Rapidbison, sorry if you misunderstood what I wrote.  I thought you were confused before.  Sorry for trying to improve your information set.  I now understand that it is more important for you to hold onto your preconceived notions.  No problem for me either way.  I did not learn math or statistics from Mr. Dancer, nor would I support anything he says if I wasn't able to understand things on my own terms.  So, I wouldn't expect anyone else to do that either.  Similarly I wouldn't believe anything simply because you wrote it.  Ironically, in the post that you are criticizing for defending Bob Dancer, I was writing about something that I don't recall him ever talking about, though I suppose he could have written about it and I just missed it.I also didn't make up the regulations that I referenced for you that require chi-squared goodness-of-fit tests.  I do realize that most may have never heard of such a test, let alone understood how to conduct one.  I wish you good health and hope you never are stricken with a disease that you believe cannot exist simply because you have never heard of it or fail to understand how it is transmitted.  I would still recommend that you seek out someone with more training in the field than you have had.There was nothing intended to belittle anyone in the previous post, mathematically or otherwise.  Though there are no direct insults in this one, you may appropriately judge this response to be more sarcastic,.  I did keep math out of it.  Certainly it is still not as mean-spirited as your response to my earlier explanation.If you want to believe that 1 out of 20 hands are not dealt randomly, I'm not certain how that helps you (unless maybe it explains why your own results have not been as positive as you would have hoped), but you are welcome to that belief.  I imagine that there are many Class II machines that are even worse than that.  If you end up with a hypothesis of how those hands are dealt on Class III machines, if not randomly, please be certain to post it to educate the rest of it.  I know now that you are uninterested in responses to your posts in areas that explain phenomena that you do not understand. However, if you truly become interested in other points of view, please let me know.  I don't read every post the way I used to do, so I might miss it.  Please be patient and have a great rest of the weekend! 
[/QUOTE]

[wildman49]

With 100k coin in,it's simply not true that you won't see a Royal most times. On average, you see one every 40,000 hands, or $50,000 coin on a quarter game.
You missed the point, true about a quarter player but Dancer does not play at that level. He plays $25 machines will only play 4000 hands to 100k. Needs ten trips to see a royal.

[billryan]

[QUOTE=billryan]
With 100k coin in,it's simply not true that you won't see a Royal most times. On average, you see one every 40,000 hands, or $50,000 coin on a quarter game.
You missed the point, true about a quarter player but Dancer does not play at that level. He plays $25 machines will only play 4000 hands to 100k. Needs ten trips to see a royal. [/QUOTE]

No, I didn't. Bob is versed enough to understand the long run.He knows for every thousand dollars, he can expect X amount of return.I've never heard him discuss short term expectations, except that they don't reflect the long run.
If your expectations per million dollars are X, then your expected return for any subset is exactly the same.

[Tedlark]

I have no problem playing a machine that may have just had a royal hit on it. I've had too many multiple royal nights, on the same machine, for something like that to stop me from playing any particular machine.

[New2vp]

I guess they use the Chi-Square test because Fisher's Exact test is still computationally too intensive and there are enough data points to converge on the results of a an exact test (give it 5 years and moore's law will set things right).There are other reasons not to use Fisher's Exact test.  Yes, the Fisher test is computationally more challenging.  Yes, the Chi-Squared test that I described requires only 52 subtractions, 52 multiplications (or squares), 52 divisions, and then 51 summations and a single comparison to the theoretical chi-squared distribution, whereas the Fisher Exact test requires computation of things like 10,000! (factorial).  Excel blows up with "only" 171!, which has "only" 309 digits.  10,000! has 35,660 digits in it.  One can get around this Excel shortcoming by using logarithms, but the real problem with the Fisher Exact test doesn't end with this level of complexity.The real problem is that in order to perform the Fisher Exact statistical test, you need to do this calculation once for the sample that you actually received to calculate its probability and then also once for every possible sample that would be judged to be more extreme than the sample that you actually saw.  Even if you can find a single, well-defined unambiguous way to do determine which samples were more extreme than the sample that you saw, there are about 7.359 x 10^137 other possible sample outcomes to calculate and make a decision as to whether they are more extreme.  We're going to need more than 5 years of Moore's law's current rate of doubling to make sense of this magnitude of calculations.  Fisher is not practical for samples of size 10,000 and most likely will not be at any time in the near future.

[rapidbison]

Simply put:

Relatively speaking, gaming regulators are using slide-rules to test continuously-generated outcomes of supercomputers.....

Apologies to all....

I stand corrected......or not.....

[asteroid]

That's where qubits come in sir - I have faith in quantum computers . . .
[QUOTE=asteroid]
I guess they use the Chi-Square test because Fisher's Exact test is still computationally too intensive and there are enough data points to converge on the results of a an exact test (give it 5 years and moore's law will set things right).There are other reasons not to use Fisher's Exact test.  Yes, the Fisher test is computationally more challenging.  Yes, the Chi-Squared test that I described requires only 52 subtractions, 52 multiplications (or squares), 52 divisions, and then 51 summations and a single comparison to the theoretical chi-squared distribution, whereas the Fisher Exact test requires computation of things like 10,000! (factorial).  Excel blows up with "only" 171!, which has "only" 309 digits.  10,000! has 35,660 digits in it.  One can get around this Excel shortcoming by using logarithms, but the real problem with the Fisher Exact test doesn't end with this level of complexity.The real problem is that in order to perform the Fisher Exact statistical test, you need to do this calculation once for the sample that you actually received to calculate its probability and then also once for every possible sample that would be judged to be more extreme than the sample that you actually saw.  Even if you can find a single, well-defined unambiguous way to do determine which samples were more extreme than the sample that you saw, there are about 7.359 x 10^137 other possible sample outcomes to calculate and make a decision as to whether they are more extreme.  We're going to need more than 5 years of Moore's law's current rate of doubling to make sense of this magnitude of calculations.  Fisher is not practical for samples of size 10,000 and most likely will not be at any time in the near future.[/QUOTE]

[New2vp]

That's where qubits come in sir - I have faith in quantum computers . . .I'm sure the future of qubits is super "qu-ool."  It is not my field of expertise, but even if it becomes possible, it is generally not worth doing 10^100 times the work to gain a decimal place or two in hypothesis testing.  If you want to know whether someone is over 6 foot tall and you can measure their height with a yardstick as 6 foot 6, it really doesn't matter if you are off by the width of an atom or two.  No sarcasm intended as that is a fairly accurate analogy  to these statistical tests.  You just want to see whether a statistic is greater than a particular measure.  If the numbers are so close as to require 5 or 6 decimal place accuracy to discern which is larger, the tests are essentially inconclusive anyway as to whether one would reject a hypothesis of randomness.I can think of other video poker problems that would benefit from faster computing.  Calculating perfect strategy for all 352,716 possible states in 10-play TDB Ultimate X Bonus Streak or calculating the exact variance for Spin Poker are a couple that come to mind.Please be sure and keep us up to date on how the future of computing can benefit our play.  I know you will have to put up with a naysayer or two here since they already don't like the amount of math involved in advanced video poker play.