Roulette Law of the third analysis

I've been doing some tests and analysis on the law of the third. I've only done a few so far but I think they're pretty informative. If anyone has insights, thinks I made a mistake, has a suggestion or request - whatever it happens to be, just let me know.

The law of the third data I have so far is somewhat interesting and I think pretty useful to an extent. Please note I used 140k numbers in chunks of 37 for this alanysis. Numbers were retrieved from random org.

Just for fun I decided to run a little test to see if there were any cycles that contained no repeaters. There were none. I'll keep the test active so whenever I add more numbers it'll check.

When do we see repeaters for the first time?

Mean Median Mode Range
2x 7 7 6 1-23
3x 20 20 20 2-36
4x 28 29 30 7-36
5x 30 31 36 12-36
6x 30 30 30 21-36
7x 30 30 30 27-34
8x 29 29 29 29-29

So only 1 8x repeater in 37 cycle for 140k numbers. At least 2 7x repeaters.

Pretty wild.

What is our occurence % for repeats?

37 spins 100% occurence
2x 100% 37 spins
3x 93% 70 spins
4x 42.69% 110 spins
5x 8.8% 160 spins
6x 1.22% 170 spins (99.87% not 100%)
7x 0.15% 200 spins (99.71% not 100%)

100% occurence is how long we'd have to wait to be assured of a win at that repeater level, but that doesn't mean every 3x goes to 4x within x spins. It just means there was a 4x present within x spins. Some of the 3x go to sleep.

I wouldn't consider anything above 5 to be very informative.

What is our average number of occurences per cycle?

0x 13.39
1x 13.79
2x 6.92
3x 2.2
4x 0.51
5x 0.091
6x 0.012
7x 0.00158
8x 0.00026

So on average we get 7 2x and 2 3x per cycle. That's across 140k spins, and I'll be adding more. I'd bet on those odds for sure.

If anyone has anything they want me to test let me know. Comments and critiques welcome

 

7X repeat took 72 spins. The 3x took past 100 units.

Nice work from up above.
But 40 spins produces on average what LOTT gives 24 non-hits. So 16 repeats.
Even Dr Sir anyones 10'330 live spins gives this. Repeats in blocks of 10 spins; 1-3-5-7=16.
Even the touted random org gives this average. Plus live German spins from Wiesbaden and air-ball as well.

 

Yes all the time.
16 repeats.
1-10; 1 repeat
11-20 5 repeats
21-30 12 repeats.
This stream is fast for repeats. But it does come the other way favouring non-hit.

 

This is from your own similar analysis with more spins? Nice to see the analysis with more data applied to it.

Why 40 spins?

So you're saying that typically the repeats occur at these spin intervals in these amounts? Interesting.

 

Not really related to the law of the third, but interesting nonetheless.

Plot showing the cumulative hit rate average as well as the actual hit rates for each number each time they hit. Also a line graph showing the cumulative hit rate over time.

The hit rate bar chart might be hard to read

 

I'm planning to use this thread to compile a bunch of research I'm doing regarding @TurboGenius and @6th-sense stuff. Whoever else I stumble across... @Nimo @Mako if they have any hints I come across as well. To be honest Mako is the reason I even started doing a lot of this research, from somewhere in the advantage of repeaters threads ( I think part 4 ) where he told @GaryG to do a bunch of research to figure it out.

More analysis:

How many repeaters per level before the next level repeater hits (2x -> 3x)
So how many 2x do we have before a 3x hits?

Repeater 2x
{'mean': 7.368253968253968,
'median': 7.0,
'mode': 6,
'mode frequency %': 10.952380952380953,
'range': (1, 23)}
Repeater 3x
{'mean': 4.30048596112311,
'median': 4.0,
'mode': 3,
'mode frequency %': 18.00755939524838,
'range': (1, 14)}
Repeater 4x
{'mean': 2.0774025974025974,
'median': 2,
'mode': 1,
'mode frequency %': 35.480519480519476,
'range': (1, 7)}
Repeater 5x
{'mean': 1.1188630490956073,
'median': 1,
'mode': 1,
'mode frequency %': 88.37209302325581,
'range': (1, 3)}
Repeater 6x
{'mean': 1, 'median': 1, 'mode': 1, 'mode frequency %': 100.0, 'range': (1, 1)}
Repeater 7x
{'mean': 1, 'median': 1, 'mode': 1, 'mode frequency %': 100.0, 'range': (1, 1)}
Repeater 8x
{'mean': 1, 'median': 1, 'mode': 1, 'mode frequency %': 100.0, 'range': (1, 1)}


Numerical pockets diff (last hit)
This one is a straight 6th-sense inspired analysis. He said in the 37 back to basics thread the numbers arent usually more than a difference of 12 numerically. And look at that...

{'mean': 12.3440329218107,
'median': 11.0,
'mode': 1,
'mode frequency %': 5.281452087007643,
'range': (0, 36)}

Repeaters v. Unhit Ratio
Another 6th-sense stat I wanted to research. He said in the same thread as above, unhits always equal repeaters by the end of the 37 spin cycle. Not exactly what I got... more like 73% on average.

calc was done: repeaters/unhits

{'mean': 0.7377481216931217,
'median': 0.7333,
'mode': 0.6667,
'mode frequency %': 8.624338624338625,
'range': (0.4286, 1.0)}

 

More interesting bet selection helpers.

 

How to combine these things? Will definitely be working an analysis on the above and below quotes.

 

Thanks for sharing your info Chrono! Are you by any chance the Chrono who used to post some strategies on the old 'Win-Max' site? Just asking because if you are, I would like to ask you something!

Regarding the Law of The Third, I did some testing / studying years ago and also found Turbo's work interesting.

Here is something I was looking at....







So it's a kind of an inverse Law of The Third because it shows that only on average, around 12 numbers out of the 37 will be overperforming and so it kind of narrows down the field for targeting bets if that's the way you want to go!

Thanks

 

Hey baccarou, no that's not me. I have used the name in the past but never for anything gambling related. Mostly in the early 2000s and 2010s.

Thanks for posting that!

I've seen that chart, I believe eugene posted it a couple years ago in the advantage of repeaters thread. I never could quite figure it out exactly.

My interpretation:

37 spins
0x - 14 numbers below expectation, not sure what the (-14) means
1x - 14 numbers at expectation
2x - 5 numbers above expectation, not sure what the +9 means
3x - 3 numbers well above expectation
4x - 1 number doing really well (hit 4 times in that cycle)

and then if I look to 111 spins
2 still have not appeared
6 is performing at expectation 3x for 3 cycles
etc..

Does that seem accurate?

I'm guessing the + and - indicate totals for above or below expectations and are just listed in the nearest cell to the cell indicating totals operating at expectation?

To me the way to use it would be to just simply ignore any numbers below expectation during bet selection, although that could be problematic as cold numbers can wake up, but if you can see in the bar graph (it's hard to see) - e.g. #16 hit rate - I posted it's not exactly reliable as to when they wake up - 1 hit is not reliable data.

 

The last video blew my mind. Absolutely insane.

 

Chrono,

In the last pic, the -14 in the first column means that 14 numbers hadn't appeared and the +9 means that there was 9 extra hits for the numbers performing above the average. When you work it all out, you are roughly getting 12 numbers performing above average.

I get your point about cold numbers waking up and that's a tricky one! Personally speaking, when playing Roulette nowadays, I like to play the dozens and will just monitor the numbers in each dozen coming above average and when I place a bet, I will just pick some from the group which are overperforming and it would probably be a good idea for me to look out for any cold ones coming out from nowhere. I think Eugene touched on that also. It's not an exact science but certainly makes for an interesting game.

 

Thinking about what you said Chrono is why Ayk's Tracker is interesting because you get those cold ones come out from nowhere. Maybe it's time to revisit all this and try and find some new insights because it's certainly an interesting concept in it's own right apart from anything else Roulette related.

 

Added as its related to the law of the third and interesting.

Yeah, I tried to do a bit of digging on cold numbers using a cumulative hit rate but it wasn't sensitive enough to catch when the cold numbers go hot. However I did have pretty good results betting splits that hadn't shown up for 60 spins, although it can get pretty sketchy. I used a neg prog. definitely wouldn't reccommmend incase the wheel is biased - you're betting on a permenantly cold number lol. A lot of times I had to take the progression like 30 to 50 levels. But it did make more than it lost in my simulations (10 players 20k rounds each, the survivers made more than the total lost).

I finally found that tracker somewhere in the forum yesterday as all links to it are no longer hosting it. I am not really sure how it works though. I also havent spent a ton of time investigating it either though.

 

Unhits and repeats are exactly the same after 37 spins....thought I would just clear that up...
As in total repeats...one number could give 2 or more repeats ....I'm not talking about individual numbers ..the syk tracker is on my back to basics thread..I think the rolling wave excel tracker is on there too...ayk tracker can be configured to 37 ..import on the step button and import however many numbers you want....to see im correct...interesting thread you are doing ...like it

 

Ah I see. I think I know what you mean. So #2 hits 2x +1 to repeater count, then #2 hits 3x so another +1 to repeater count? Etc..

I posted a reply to @baccarou above, not sure if you can see it... For some reason it says its moderated and all the text is whited out?

Horseraces

I've got some horserace graphs here. Been trying to find a pattern, tried so many different ways of looking at it. Yes the obvious pattern that the spins gaps are less than 35. I've tried looking at it as initial position to see if they keep rank order, I have animated graphs of these, etc.. There's definitely some other way to look at it that I'm not considering. Clearly though, when you see these animated in action the most common situation is for it to fill out fairly evenly. It's almost like it the different levels take turns.... randomly. Odd but unpredictable so far. Definitely the most obvious finding is that the last numbers out usually REMAIN last for quite some time.

These are just the standard horseraces we can see in turbos excel photos, I made them fancy to help show any patterns. I have a couple variations including replacing the pocket number with the number in which it came out. A form of indexing to see if order is preserved... Not really.

In these graphs you can see in # 5 6 7 they are 37 spin cycles, but the first, second and third graphed separately. #8 is all 3 together in order they came out and the rest are obvious if you read the titles of the graph. It's all from the same game.



In these next ones, they are all different games of 10 cycles. Just for fun. If anyone sees anything they better tell me! lol



Going to be running more tests on the horserace. Checking top #s from each doz, double streets as well maybe. Testing dropping numbers out etc..

 

Thumbs Up

 

For some reason my other 2 replies are whited out and moderated - no idea why. I copy and pasted them here again.

Keeping this as it's interesting.

Yeah, I tried to do a bit of digging on cold numbers using a cumulative hit rate but it wasn't sensitive enough to catch when the cold numbers go hot. However I did have pretty good results betting splits that hadn't shown up for 60 spins, although it can get pretty sketchy. I used a neg prog. definitely wouldn't reccommmend incase the wheel is biased - you're betting on a permenantly cold number lol. A lot of times I had to take the progression like 30 to 50 levels. But it did make more than it lost in my simulations (10 players 20k rounds each, the survivers made more than the total lost).

I finally found that tracker somewhere in the forum yesterday as all links to it are no longer hosting it. I am not really sure how it works though. I also havent spent a ton of time investigating it either though.

Ah I see. I think I know what you mean. So #2 hits 2x +1 to repeater count, then #2 hits 3x so another +1 to repeater count? Etc..

I posted a reply to @baccarou above, not sure if you can see it... For some reason it says its moderated and all the text is whited out?

Horseraces

I've got some horserace graphs here. Been trying to find a pattern, tried so many different ways of looking at it. Yes the obvious pattern that the spins gaps are less than 35. I've tried looking at it as initial position to see if they keep rank order, I have animated graphs of these, etc.. There's definitely some other way to look at it that I'm not considering. Clearly though, when you see these animated in action the most common situation is for it to fill out fairly evenly. It's almost like it the different levels take turns.... randomly. Odd but unpredictable so far. Definitely the most obvious finding is that the last numbers out usually REMAIN last for quite some time.

These are just the standard horseraces we can see in turbos excel photos, I made them fancy to help show any patterns. I have a couple variations including replacing the pocket number with the number in which it came out. A form of indexing to see if order is preserved... Not really.

In these graphs you can see in # 5 6 7 they are 37 spin cycles, but the first, second and third graphed separately. #8 is all 3 together in order they came out and the rest are obvious if you read the titles of the graph. It's all from the same game.



In these next ones, they are all different games of 10 cycles. Just for fun. If anyone sees anything they better tell me! lol



Going to be running more tests on the horserace. Checking top #s from each doz, double streets as well maybe. Testing dropping numbers out etc..