Roulette Law of the third analysis

This was posted from rrbb so we split things in two, the text explain what happening but as you said yes!

 

Ok great. Thank you.

Do you have a link or something I could read more about this?

 

https://www.rouletteforum .cc/index.php?topic=19110.0

Lets try again

 

Very interesting... I like the idea. It's as a lot of people say, go with the flow - let the board tell you where to bet. I just finished going through turbos examples on his thread informational: what are your chances of winning. Similar idea, also ended in positive after 3700 spins with the same profit as him at the same time.

Very, very weird.

Thank you for pointing this out to me. I'm going to investigate further. Any comments are always appreciated!

 

Funny thing about Turbos 'informational what are you chances of winning' example thread is it is similar to the prisoner problem. I am thinking the birthday paradox and prisoner problem are related when it comes to roulette... Just a thought.

 

The law of the third is not what you think it is. It merely demonstrates the most likely way in which numbers can be distributed over a series of spins. For example, it's very unlikely that you will see every number hit exactly once over 38 spins, since there are exponentially more ways for them to not hit exactly once. It does not tell you the numbers that are likely to start or stop hitting on the next series of spins.
Furthermore, if you do the math, the most common distribution merely proves that the house edge is intact and insurmountable.

There is a different way that you can use the law of the third. You can use it to exploit wheels that aren't random. Namely, live wheels that spin only in one direction, machines, or by segregating data by spin direction. Look for results that regularly don't quite conform to the most likely distribution. Use the law of third as a kind of wheel fitness test. You should also use chi square testing and data correlations as well.

Stop testing puny amounts of spins. 37 spins is utterly meaningless. Increase the sample sizes into the hundreds and thousands. Correlate data sets.

Do NOT introduce noise into your data in the form of streets, splits, corners, and columns. If you're going to find an edge, you do NOT want to dilute it with random numbers. Please stop with the street nonsense, you're just wasting your time.

 

Dr.Sir Anyone if I post 500# from a live table and 500# from .org could you tell the difference?

 

Better than probability would expect. By the time you reach a very significant amount of spins 50k, you can rather easily differentiate which is an RNG and which is a live wheel. It's not just the chi square that stands out, it's also the locations of the weakest numbers in relation to the strongest numbers. Every live wheel is biased to some degree. I have such spin samples.

 

I'm testing a 37 spin horizon on 240k total numbers. The horizon as you call it seems to be an arbitrary number anyway - I'm just using 37 as that's the most commonly used range. That way I can verify or disprove other claims being made.

The horizon I've defined in my code is 1 variable, at any point I can change the number and have my code spit out everything I've posted here for any betting horizon.

I know that the law of the thirds is just a byproduct of the birthday paradox, and not a magical prediction mechanism.

I completely agree that betting anything other than a straight-up is not a 'best-practice'. The payouts are worse - rounded down from what their true comparison against a straight-up should be, give less room for error, and in all cases (besides splits and streets) don't include the 0 further reducing any edge the player has.

The more numbers a bet includes the more the casino wants you to place that bet.

So how common are exploitable biased wheels?

 

The only reason Real = DSAA is saying that is coz its incompatible with his style of play -- once the alleged advantageous interval of the wheel has been identified, the only way he can exploit that is by placing a set of contiguous SU-numbers as a spin-bet.

@Chrono once you account for, beside the house edge paid, also the probability .. you'll realize that what I call the 'raw credits' taking shape into the reality are just the same & either way you play (EC, ...,SU), the perks you get by opting for any equalizes out.

Let's take into the example 18ⁿ = numbers played -- you can play that in many & various ways, but for the purposes of this let's focus on extremes, playing SU for 18-spins versus all 18ⁿ in a single spin.

True, the former gives you the advantage of paying less house edge out .. but at the cost of lower probability of hit, reduced by almost ≈25% -- (48.649/38.932= 1.24958902702).

In other words, you'd have to play 24-25 numbers on average, as SU over 24-25 spins, investing more ≈36% more money in in order to get the same probability of a hit return -- (24.5/18= 1.36111111111).

As mentioned previously, you may play these 18ⁿ in many & various way ... eg. only one payout-type & using DS, as a ≈mid-type of the payout (1:1 ... 5:1 ... 35:1) between these two extremes, & thus have these two parameters (he vs probability) somewhat balanced out by default; or you may use a parachute-based system, that shifts between the payouts, having as the pivotal- of garry-point parameter the exposition = other parameters aligning around that, basically placing the bets in most cost-effective way based on the number of numbers played ≈combined-payout (not the default ones offered by casinos eg. 2Qs or two-corners, 7:1).

Now, we are back at raw credits & how do you want to materialize those .. ever played an rpg, where you distribute the raw credits about the various parameters given (strenght, stamina, speed,magic), shaping a character.

Makes sense?

PS:
Btw, @TwoUp has the same thing as above worded a bit differently, more of math-themed in one of his posts -- try searching using the keywords ... "house edge" & "su".. "50%", "24 spins", "25 spins".

 

So DSAA , this is from a live wheel I follow where i am, and i mainly collect 500 spin from that wheel..
This is what it look cycles of 111..

 

Here's the average expected times of when you can expect each position in the horserace to show up. It works pretty good across 200k spins actually. Sometimes it's really accurate, but it depends on the game. I have more but I'm still working on it.

This is built on a 10 cycle horizon, averaged across 120k records. It's computationally intensive to calculate so I didn't run it against the full 240k I have.

Yes it's hard to read, it's also better than nothing considering the average game will tend to play out this way.

How to read it: (SPIN #, VARIANCE, (MIN SPIN, MAX SPIN))

Variance is a % and shows +/- the variance of the spin # based off the min and max spin numbers. The range (min, max) and variance are basically the same - two different ways of looking at it.

This is all for straight-up bets

Example:

Level 1 - one hitters
1: [(1, 0.0, (1, 1)), ...]
First number out on spin 1 on average, has 0 variance in the range and the range it spin 1 min, spin 1 max...... (obviously)

Level 3 - 3 hits

3: [(21, 72.66, (3, 52)), ...]
Third number to repeat for the first time happens on spin 21 on average, there's a +/- 72.66% variation from this average, on average, and the range we get the first 3peat is between 3 and 52.

13: [(246, 45.14, (97, 350)), (280, 33.74, (158, 368))],
There are 2 numbers that have hit at this level
One hit at spin 246, the other at 280.
Variance was +/- 45.14% & 33.74% respectively
The hits happened from spins 97 - 350 and 158 - 368 respectively

So treat each '(21, 72.66, (3, 52))' as if it were the numbers in the horse race.

All you really need is the first number of each (21, 72.66, (3, 52)) to build an excel spread sheet model.

@baccarou If you're still working on the horse race stuff you might find this helpful

@atrox23 I think I'm seeing what you mean by only looking at unhits and uniques - considering each level of the horse race it's own game, with uniques and unhits for that level?

1: [(1, 0.0, (1, 1)), (2, 30.0, (2, 5)), (3, 28.57, (3, 7)), (4, 27.78, (4, 9)), (5, 25.0, (5, 10)), (6, 26.92, (6, 13)), (8, 29.58, (7, 15)), (9, 31.87, (8, 19)), (10, 30.0, (9, 20)), (11, 28.35, (10, 21)), (13, 30.61, (11, 24)), (14, 30.22, (12, 26)), (16, 30.8, (13, 28)), (17, 29.51, (14, 29)), (19, 31.74, (15, 33)), (21, 34.27, (16, 38)), (22, 35.17, (17, 42)), (24, 36.97, (18, 47)), (26, 38.46, (19, 52)), (28, 40.15, (20, 58)), (30, 40.41, (21, 61)), (33, 40.48, (22, 63)), (35, 41.02, (23, 67)), (38, 42.75, (24, 74)), (41, 43.23, (25, 78)), (44, 43.63, (26, 82)), (47, 43.95, (27, 86)), (51, 42.57, (30, 91)), (55, 45.12, (31, 103)), (60, 47.29, (33, 119)), (65, 48.05, (34, 126)), (71, 48.16, (36, 134)), (79, 49.08, (38, 147)), (88, 52.96, (40, 181)), (101, 53.02, (47, 213)), (119, 57.17, (51, 278))],

2: [(8, 73.21, (2, 28)), (13, 62.2, (4, 29)), (17, 57.35, (6, 34)), (20, 52.97, (8, 37)), (23, 48.0, (10, 38)), (26, 44.05, (13, 42)), (29, 40.89, (16, 46)), (31, 40.95, (17, 49)), (34, 40.2, (18, 51)), (37, 34.66, (23, 54)), (40, 37.21, (24, 61)), (42, 36.04, (27, 66)), (45, 34.69, (29, 68)), (48, 32.86, (32, 71)), (51, 33.55, (34, 77)), (53, 31.97, (37, 80)), (56, 31.44, (39, 83)), (59, 30.5, (42, 87)), (63, 28.89, (45, 89)), (66, 30.05, (46, 94)), (69, 30.33, (49, 101)), (73, 31.32, (50, 106)), (76, 30.55, (54, 112)), (80, 32.01, (55, 119)), (84, 31.28, (59, 125)), (89, 31.4, (63, 134)), (94, 32.14, (64, 139)), (99, 32.88, (68, 151)), (104, 34.25, (70, 162)), (111, 34.59, (74, 173)), (117, 34.7, (78, 183)), (126, 37.46, (82, 210)), (135, 39.63, (88, 243)), (146, 41.21, (93, 271)), (162, 43.06, (99, 307)), (184, 47.35, (102, 369))],

3: [(21, 72.66, (3, 52)), (29, 64.22, (6, 57)), (36, 52.44, (13, 61)), (41, 48.21, (17, 66)), (45, 43.87, (22, 71)), (50, 39.75, (26, 73)), (54, 36.68, (31, 78)), (58, 35.71, (35, 85)), (62, 36.02, (38, 93)), (65, 32.71, (43, 95)), (69, 29.65, (48, 97)), (73, 30.66, (50, 104)), (77, 28.96, (55, 109)), (80, 28.98, (57, 113)), (84, 27.48, (61, 116)), (88, 28.15, (64, 124)), (92, 27.38, (68, 129)), (95, 27.71, (70, 134)), (99, 27.52, (74, 141)), (104, 25.92, (80, 146)), (108, 26.04, (82, 150)), (112, 26.38, (85, 157)), (117, 26.5, (88, 163)), (122, 26.2, (92, 169)), (127, 27.3, (97, 184)), (132, 27.58, (99, 189)), (138, 27.29, (104, 197)), (144, 26.86, (108, 202)), (151, 27.96, (111, 214)), (158, 28.81, (113, 223)), (166, 32.08, (117, 254)), (176, 31.06, (124, 261)), (187, 32.78, (130, 288)), (200, 35.23, (136, 325)), (218, 37.38, (141, 360))],

4: [(37, 71.18, (4, 79)), (49, 57.57, (13, 84)), (57, 46.75, (25, 91)), (64, 45.47, (28, 98)), (70, 37.83, (39, 102)), (76, 36.19, (44, 109)), (81, 34.21, (48, 112)), (86, 29.21, (58, 116)), (91, 28.09, (63, 122)), (96, 27.31, (67, 127)), (100, 26.33, (71, 131)), (105, 25.87, (74, 135)), (109, 26.43, (79, 146)), (114, 25.41, (84, 151)), (118, 23.98, (89, 154)), (123, 23.52, (93, 159)), (127, 23.23, (99, 168)), (132, 22.77, (102, 171)), (136, 22.46, (107, 178)), (141, 24.05, (109, 189)), (146, 22.94, (114, 192)), (151, 22.6, (118, 197)), (157, 23.59, (122, 209)), (162, 22.99, (128, 216)), (168, 23.84, (131, 226)), (174, 24.6, (133, 234)), (181, 25.84, (140, 255)), (188, 25.94, (142, 259)), (195, 26.28, (144, 265)), (203, 25.16, (154, 275)), (213, 25.8, (158, 287)), (223, 29.05, (162, 322)), (235, 28.54, (172, 337))],

5: [(56, 68.52, (7, 111)), (71, 51.52, (25, 115)), (81, 47.32, (32, 123)), (90, 40.12, (45, 129)), (97, 34.18, (58, 135)), (104, 33.74, (60, 139)), (110, 28.36, (73, 143)), (115, 26.95, (81, 152)), (121, 25.68, (86, 156)), (126, 25.64, (90, 163)), (132, 26.01, (91, 167)), (137, 26.69, (93, 174)), (142, 23.74, (107, 184)), (147, 22.0, (115, 189)), (152, 21.27, (119, 192)), (158, 20.97, (123, 197)), (163, 21.43, (129, 209)), (168, 20.16, (137, 215)), (173, 20.4, (140, 221)), (179, 20.35, (144, 227)), (184, 20.91, (147, 235)), (190, 21.07, (152, 244)), (196, 20.49, (158, 250)), (202, 20.75, (162, 258)), (209, 21.87, (166, 272)), (216, 21.06, (175, 281)), (223, 20.74, (179, 285)), (231, 20.92, (182, 291)), (239, 21.62, (188, 306)), (248, 21.61, (196, 319)), (258, 24.67, (198, 349)), (270, 25.45, (205, 369))],

6: [(76, 66.25, (12, 147)), (94, 52.06, (32, 152)), (106, 44.29, (50, 165)), (116, 38.5, (64, 171)), (125, 34.67, (73, 173)), (132, 31.01, (82, 174)), (139, 28.08, (92, 179)), (146, 26.92, (100, 188)), (152, 24.23, (110, 192)), (158, 23.82, (114, 197)), (164, 24.49, (119, 209)), (170, 23.11, (127, 215)), (176, 22.99, (129, 218)), (181, 23.05, (131, 222)), (187, 21.11, (143, 230)), (193, 21.02, (149, 239)), (199, 19.26, (161, 247)), (204, 18.92, (165, 251)), (210, 20.38, (168, 265)), (216, 20.77, (172, 274)), (222, 19.87, (181, 282)), (229, 19.71, (185, 287)), (235, 19.96, (189, 295)), (242, 20.8, (195, 311)), (249, 19.48, (206, 318)), (257, 19.89, (211, 329)), (265, 20.99, (214, 343)), (273, 20.82, (220, 351)), (282, 20.26, (227, 357))],

7: [(98, 61.98, (19, 173)), (119, 47.67, (45, 178)), (133, 40.94, (62, 186)), (144, 37.35, (73, 193)), (153, 32.58, (91, 203)), (161, 29.5, (103, 209)), (169, 27.49, (111, 213)), (176, 26.76, (117, 220)), (183, 24.18, (128, 224)), (190, 22.38, (138, 230)), (197, 23.17, (143, 243)), (203, 20.38, (161, 254)), (209, 20.02, (165, 258)), (216, 18.43, (177, 266)), (222, 17.82, (183, 271)), (228, 17.98, (185, 275)), (235, 18.27, (189, 283)), (241, 17.02, (199, 289)), (247, 17.94, (202, 300)), (254, 17.22, (209, 305)), (260, 17.5, (213, 313)), (267, 18.53, (215, 324)), (274, 17.43, (225, 330)), (281, 16.49, (235, 337)), (289, 17.25, (241, 352)), (297, 17.73, (243, 359))],

8: [(121, 57.98, (27, 196)), (144, 41.95, (67, 207)), (160, 35.6, (87, 215)), (172, 31.08, (105, 224)), (182, 29.23, (119, 239)), (191, 27.15, (130, 246)), (200, 24.18, (142, 248)), (208, 23.8, (148, 256)), (215, 21.82, (157, 258)), (222, 20.3, (172, 271)), (230, 20.29, (175, 276)), (237, 19.16, (184, 282)), (243, 18.76, (194, 294)), (250, 17.25, (204, 298)), (257, 15.96, (214, 303)), (264, 15.84, (221, 312)), (270, 17.24, (228, 333)), (277, 16.6, (233, 335)), (284, 16.98, (235, 341)), (291, 17.3, (238, 348)), (298, 15.46, (253, 354)), (305, 15.65, (259, 364)), (313, 15.89, (261, 369))],

9: [(144, 52.24, (47, 229)), (170, 39.73, (85, 241)), (187, 34.78, (100, 243)), (201, 30.24, (122, 255)), (212, 27.22, (139, 265)), (222, 24.21, (156, 273)), (231, 23.26, (166, 283)), (239, 20.99, (180, 289)), (248, 20.23, (192, 302)), (255, 20.7, (196, 312)), (263, 19.61, (206, 319)), (270, 18.39, (215, 323)), (278, 16.84, (226, 327)), (285, 16.63, (237, 341)), (292, 16.17, (241, 343)), (299, 17.04, (245, 356)), (306, 17.05, (249, 362)), (314, 16.57, (256, 368))],

10: [(169, 50.12, (60, 263)), (197, 38.02, (101, 271)), (215, 31.14, (131, 280)), (230, 29.93, (138, 287)), (242, 24.07, (169, 295)), (252, 23.63, (176, 304)), (262, 21.84, (193, 317)), (272, 21.52, (200, 326)), (280, 19.03, (218, 333)), (289, 18.89, (226, 344)), (297, 17.48, (241, 354)), (305, 16.46, (249, 357)), (313, 16.78, (254, 367))],

11: [(194, 47.13, (74, 287)), (224, 38.66, (115, 314)), (244, 32.91, (139, 316)), (259, 28.7, (161, 322)), (272, 25.2, (188, 338)), (283, 24.29, (195, 343)), (294, 20.99, (219, 352)), (304, 20.25, (231, 364)), (313, 19.36, (240, 370))],

12: [(220, 45.64, (87, 318)), (252, 35.69, (134, 334)), (272, 31.14, (157, 340)), (288, 26.35, (197, 365))],

13: [(246, 45.14, (97, 350)), (280, 33.74, (158, 368))],

14: [(272, 41.92, (116, 370))],
15: [(297, 38.82, (125, 370))],
16: [(316, 34.83, (142, 370))],
17: [(328, 30.37, (166, 370))],
18: [(335, 26.82, (187, 370))],
19: [(340, 25.38, (195, 370))],
20: [(342, 24.84, (198, 370))],
21: [(344, 21.68, (219, 370))],
22: [(346, 14.51, (268, 370))],
23: [(348, 12.74, (280, 370))],
24: [(348, 11.88, (286, 370))],
25: [(360, 4.55, (337, 370))]

Personally I think there's probably about 3 or 4 different ways the game tends to play out - sometimes we get a very average game like above, sometimes a very wide (lots of unhits for each level before moving to next level) or below average game, and an above average game where you get one or two numbers just ripping through the levels fast and early.

Interesting thing is the spin # variance decreases in the middle parts of the levels, with it being higher towards the beginning and ends...

The funny thing is every horse race table I've generated there's always a profitable bet at any point - horizontally or vertically. Of course that's after the game is over... Anyway this was a fun little analysis.

 

Just a note to add on, the spin # variance is low towards the end of the cycle due only in part to a lack of data available in that cycle. It's not because the accuracy is increasing. 370 spins have numbers that can only go into so many places. 740 spins would show a relative variance at the same point as 185 spins would show in the 370 spin cycle. The interesting data is the fact that the variance is decreasing in the middle of the levels.

Less variance = higher predictability in my opinion.

 

Thanks very much for this Chronos! I was hoping you would do something for this because it's kind of like the missing piece for all the stats I have on the LOTT. I will have a good look at it after dinner!

 

No problem buddy.

The simple thing I'm missing from the horserace is driving me crazy, lol!

 

Personally, my thoughts on the horserace are that it's not a system or way to be bet on.

It may not even be about repeaters.

But I do think it ties into his geiger counter video he kept referring to.

A lot of the things he says point to cold numbers becoming hot, and hot numbers becoming cold. The horserace is probably a way to conceptualize the 'geiger counter' he mentions.

 

My data shows clearly what LOTT says, i am not mixing it with turbo thing !
From the first cycle for example :
33 x1 from them appeared 27
27 x2 from them appeared 22
22 x3 from them appeared 13 etc...

 

Chrono and all!

I had an idea tonight regarding this racehorse scenario which kind of fits in nicely with the LOTT!

Let's assign numbers from 1-12 1,2,3,4,5,6,7,8,9,10,11,12.
Assign numbers from 13-24 1,2,3,4,5,6,7,8,9,10,11,12.
Assign numbers 25-36 1,2,3,4,5,6,7,8,9,10,11,12.

This way we only work with 12 numbers in the racehorse scorecard instead of 36.
We know because of the LOTT that there will be hot numbers and cold numbers.
We also know towards the back end of a cycle, we will start getting more repeats.

In the picture below, you can see that we get a little purple patch of repeats from around spin 29 to 35.
On spins 31 and 33, we see that the 12 has come out for the 6th time (spin 31) and that the 3 has come out for the 5th time (spin 33)
You can also see that the actual number 12 has hit for the 3rd time (spin 31) and the actual number 15 has hit for the 3rd time (spin 33)

If playing for a 12 which is either number 12, 24 or 36, you can see that the 12 and 24 both hit twice and the 36 only once just before spin 31 and so you would likely only play the 12 and 24.

If playing for the 3 which is either number 3,15 or 27, you can see that the 3 and 15 both hit twice and the 27 is yet to appear just before spin 33 and so you would likely only play the 3 and 15.

Looking to the far right where I have my racehorse scorecard, you can see that I am only playing the top 3 and it is pretty much consistent with the 1,3,5 and 12 dominant. This might be a nice way to break it all down.

 

Chrono based on your data y have to run a bet now, that will bet what you where to bet and see. There is no point testing millions off spins, if something works it has a reason...

 

Chrono
Have a look at reply 54 in if 16 repeats at roulette forum.
You'll see how many times could bet a non-hit number and how long seen it take to hit.
like betting last 9 non-hit numbers. bet for 14 spins and has a max of 23 spins to hit.
This data is very old and was used on FOBT's in UK bookmakers like Corals.
Had to use .20P unit as max bet was £100.
Not available now as UK Gov has made FOBT's max £2 bet.