As far as 'live' roulette is concerned (a wheel, a ball and random outcomes) we're told that in a spin cycle of 37 only about 24 numbers will appear.

And of those 24, around 13/14 (yes, about a third)will normally repeat --some twice, some even 3 or 4 times.


But does that Law Of The Third apply to man-made, casino software?

Is all RNG software deliberately written to ensure the Law Of The Third actually occurs? Or only some RNG software?

No guesses or speculation, please. Only answer if you know, or if you can point to an authoritative, credible link.

Nearly 100 views and nobody has an answer. I'm not altogether surprised by that. 

I suspect the RNG software is NOT coded to replicate the Law. 

It would have been good if that suspicion could have been confirmed or disproved.

esoito,


I don't know why such a big deal is made of the 'law of the third'. There is nothing special about it and there are numerous other 'laws' that will manifest given that outcomes are random and independent. By 'random' I mean that each number (or group of numbers) has an equal likelihood of occurring.


It isn't necessary to 'code in' the law of third to casino software because it occurs naturally; all you need to do is make sure that outcomes are random and independent and the law will manifest inevitably as a natural consequence. There is no mystery about it; it works because when you start with a relatively large set of numbers (37 in this case) you are more likely to get different numbers on successive spins early on, but later, when several numbers have been selected, you are increasingly likely to get a repeat because there are then more selected numbers than unselected numbers remaining - it's simple logic.


Try it yourself in a spreadsheet (or better still, using a simple programming language, perhaps Victor's new 'vASIC' assuming it has a built-in RNG). You don't need any authoritative links for any of this because you can prove or disprove it for yourself, and isn't that the best way to prove anything?

An answer based on data that I collected playing BVNZ. There are 67000+ spins. To me it appeared similar to data I collected from B&M casinos. As you can see most of the 37 spin sessions had 21-26 unique numbers. So, I believe law of third works to a certain extent in RNG as well.

[csv=,]
Total unique numbers in 37 spins,Number of times
18,6
19,33
20,77
21,191
22,358
23,391
24,347
25,229
26,103
27,39
28,34
29,10
30,1
[/csv]

Well done, Pockets.     Very helpful and most interesting. thank you for taking the time and trouble to do that.

If Mike is right then it's not specifically coded but occurs as a natural phenomenon.

It's the specific ratio 1:3 that I found interesting. Why is it not 1:1...or 1:2...or 1:4? Why 1:3?

Aren't numbers fascinating?

Thanks to both of you.

Quote from: esoito on November 13, 2013, 09:37:07 AM
It's the specific ratio 1:3 that I found interesting. Why is it not 1:1...or 1:2...or 1:4? Why 1:3?
Aren't numbers fascinating?

Indeed they are. I will try to answer your question in simple terms in case you are interested. First lets start with Even chances. Red and black. So in two spins, the possible outcomes are RR, RB, BR, BB. Now if you see here, out of 4 possible outcomes, 2 are repeat outcomes and 2 are unique outcomes. So if you define a law of third equivalent for ECs, it will be a law of halves as the outcome probabilities indicate it.

Now, lets take dozens 1,2,3 and three spins.  There are 27 possible outcomes out of which only 6 outcomes will have unique dozens. So as you can see you get a greater chance of a repeat dozen in 3 spins than you would find in case of ECs and this split will determine your question of whether it is 1:1 or 1:2 or 1:3 and so on.

Now all you need to do is extrapolate this to 37 outcomes and 37 spins to see why it is law of third. I have put things in a simple manner so that it is understandable. Perhaps our friend Bayes who is missing for some time now might be able to give a better answer. Hope this helps Max.

Now whether anyone can exploit this is definitely a questionable and we can argue for a life time and many more years to come

I have previously simulated nearly 200,000 cycles of 37spins, using Bayes's one million actuals, a million or so from random.org and the rest from Excel. Excel's RNG is far from perfect and is susceptible to producing results with a pretty large range of fluctuation. Though, we can probably assume that's because more than two thirds of the simulation was done with Excel numbers. I can't vouch for a casino RNG but repeat numbers are just the nature of the game. If it's truly random, there will be a zero or small margin of error when compared to a real wheel; or better a TRNG like random.org. In my opinion, that will be the case in nearly all tests.

Against a casino RNG the margin of error would only be significant after the fact, i.e. after you have played against it and most likely lost, then you can begin to think about manipulation. But you can't ever really be certain of any degree of manipluation unless a bet selection has a real egde. Real meaning it can be proved with a simulation. If it relys on "guessing" or "experience", there is no edge, and any so-called maniplation of the RNG could simply be down to the perplexity of the player.

Ah, Pockets...

Thank you for taking the trouble to post your explanation. 

I hope others -- especially RNG players -- found that as informative as I did.


And more thanks to Number 6 for those revelations.  [As an aside, BAYES has been AWOL for some time now, and there are concerns for his well-being. None of his three email addresses are responding. So if anyone can throw some light, please PM me. ]


See, the main reason I posed the original question was because so many seem to play RNG rather than the real game.

And if they are using some system or other based on the LOTT then they could be losing before they even place a bet, if the LOTT-phenomenon is absent from RNG !!

Hence my original question, and I asked for an authoritative reply so the thread didn't fill up with 'speculative noise'...LOL 

I had rather hoped a former casino programmer -- who would likely know for certain -- might have already addressed the matter somewhere on the net....