RNG & DEAL/DRAW button timing

[MikeA]

Just went through 3,200 credits without a 4-of-a-kind on my Memorial weekend casino trip.What game or games do you seek?  I don't recall though you have probably posted.  Regardless, that is way below average!  I generally expect one every 400-500 hands with DB and DDB.

[Benford's Law]

Mike,     It was 25-cent AAM (All-American) with 4-of-a-kinds paying 200 (100.72%).  Just fortunate that the 3,200 credits were lost on a low-denomination machine.  It took only 4-1/2 hours to go through these 3,200 credits playing 6 different machines, so the lower hands (FH and lower) were scarce, too.  Pretty wicked for a game with only a 26 variance rating.~Benford's Law

[shadowman]

In AA you should see quads every 444 hands. If I assume a paying rate of at least 600 hands/hr then you would have gone through over 6 cycles without a quad. This is pretty close to the worst case I have ran into(except for my 13 cycle WRF drought in deuces).
 
I've found variance to be somewhat unrealiable in determining playing time ... Of course, 26 would be considered high if VP didn't have other games 40 and above ...  I prefer to look at the amount of payback in commonly occuring hands (FH and below). With AA that number is ~86% which is not bad but not all that great either. A JOB progressive (5200 credits) has a variance of 34 but over 90% of the payback is on these kind of hands. I think this demonstrates why variance isn't always that useful.

[Benford's Law]

Shadowman,        I see what you are saying in your last post and two things come to mind:       a)  On the JOB progressive with a 5,200 payout for the royal the variance is driven higher by the fact the player must sacrifice some lower paying hands for a shot at the royal.  90% of the payout is no longer driven by the lower-paying hands in this case.  I don't play progressives too often and I haven't taken a look at Zamzow's strategy master or Dancer's program as I write this down, so I am responding here off "the top of my head".  In fact, I don't know for sure how variance is exactly calculated...I only use it as a quick and simple guide to how severely my bankroll will fluctuate on a given game.       b)  While it is true that I went through 5-6 cycles without a 4-of-a-kind I was also getting smacked by the absence of lower hands in that session, too.  I've had a few sessions where I went through 5 cycles without a quad at 25c AAM but they historically have only cost me $200-$500 instead of $800 (depending on my luck with the 40-coin and lower hands). ~Benford's Law

[shadowman]

My value of "over 90%" is directly from winpoker (90.78). The effect of progressives is not really all that significant on the lower paying hands as the value is 91.10 without the RF.
I also hate it when the mid and low paying hands take a long vacation. Even when your getting a few quads they get eaten away quickly. I've seen 800 credits disappear in 5 minutes at 3-play JOB. Like I said before the variance of VP is moderate no matter what the game you are playing. JOB only seems low by comparison to other VP games. I still prefer looking at the payback of these moderate hands over using variance. Especially when playing progressives.

[Benford's Law]

Shadowman,        Please let me neurotically get the last word in by saying I see your point with checking the payoffs of hands below 4-kind in non-wild card games over variance.  I still wish I knew how variance is calculated, though.~Benford's Law

[shadowman]

I had the formula for variance at one time but must have pitched it. In general it is the sum of the distance from the mean squared, divided by the number of elements. SUM(x(i)-mean)**2/N. The mean of a VP game is the payback. So, my best guess is you take each paytable entry and subtract the mean (800-.9954)**2 for the RF element of JOB, then multiply by the probability of RF. Do this for each entry and the divide by the total entries.

[faygo]

I had the formula for variance at one time but must have pitched it. In general it is the sum of the distance from the mean squared, divided by the number of elements. SUM(x(i)-mean)**2/N. The mean of a VP game is the payback. So, my best guess is you take each paytable entry and subtract the mean (800-.9954)**2 for the RF element of JOB, then multiply by the probability of RF. Do this for each entry and the divide by the total entries.
 
 OK! OK! That does it.
You guys made me read about the Black Swan, Occam's Razor, and Mr. Benford"s Law. Being Math Challenged my head is ready to explode.
 
On a serious note, you programmers out there, how would Benford"s law  affect RNG programming? From the little I read, it might make a case for (I'm going to say it) Hot and Cold cycles due to programming.
 
I'll wait for my beating.
 
Faygo
 

[shadowman]

Benford's Law only applies to the first digit of a set of numbers. It is not relevant to VP machine RNGs. The key element of a valid RNG is that all numbers have an EQUAL opportunity to occur. Therefore, by definition, Benford's law does not apply.
 
It turns out that Benford's law can be used to determine if a set of numbers has been computer generated. If the set of numbers in question (that should follow Benford's Law) has an even distribution then it was probably faked.

[faygo]

Benford's Law only applies to the first digit of a set of numbers. It is not relevant to VP machine RNGs. The key element of a valid RNG is that all numbers have an EQUAL opportunity to occur. Therefore, by definition, Benford's law does not apply.
 
It turns out that Benford's law can be used to determine if a set of numbers has been computer generated. If the set of numbers in question (that should follow Benford's Law) has an even distribution then it was probably faked.


I think I understand it now Shadowman. Thanks
Now let me ask, If as I have read, the RNG is a Pseudo-RNG relying on a seed number and looking at the last number generated to produce the next Random number would that cause these streaks? Or would this only apply to Slot machines?
My problem I guess is just like understanding Electricity: Can,t see it. (Unless it tries to escape)
 
Faygo