John Legend's Pattern Breaker

[falkor]

John Legend's system could be the most widely read system on this board; check out this topic in case you are unaware of it's reputation:
link:://:.rouletteforum.cc/index.php?topic=4655.0

I am going to analyse Pattern Breaker to see if - after tracking - there's a pattern to the next 3 spins based on the last pattern and the missing 8th pattern indicating that it's less likely to appear compared to any of the other 7 patterns.

I've got the basic engine running (see attached); before I begin analysis does this seem correct and can you spot any bugs?

[falkor]

Total wins: 274, Total losses: 42, Max wins in a row: 27, Max losses in a row: 3, Tally: 42, 47, 41, 30, 43, 33, 42, 38

(That's from specifically testing missing pattern HHH and then betting against it (LLL) from 320 sets over 100,000 spins)

Total won = 548
Total lost = 588

A 4-step progression would have saved it. HHH came 42 times and lost, but HLL only came 30 times; LHL 33 times.

Is the above comparable to dividing the board into 8 and betting 7/8? I wonder what's the most losses in a row this could go up to when playing all patterns over a 1 million spins?

Next test: HLH

[falkor]

Yep - it's equivalent to this:
   Board divided into 8   Pattern Breaker
Progression 1   14   14
Progression 2   112   112
Progression 3   896   896
Progression 4   7168   7168

[ego]

-

This is the result got testing pattern breaker ...
The longest strike i got was 23 in a row and the largest losses 2 in a row.

2014 09 27

WL
WWWWWWWWL
LLW
WWWWWWL

2014 09 26

WWL
LW
LW

2014 09 25

WWL
WWWWWL
WWL
LLW
WWWWWWL

2014 09 24

LW
WL
WWWL
LLW

2014 09 23

LW
LW
WWL
WWWL

2014 09 22

WWL
LW
WL
LLL
LLW

2014 09 21

WL
LLL
WL
WWWL
LW

2014 09 20

WL
LW
WL
WL

2014 09 19

LW
LW
LW
WL
WL

2014 09 18

LLL
LLL
WWWL
WWL

2014 09 17

LLW
WL
WL



I test this using Regression Up & Pull.
So if i won the first bet i continue to bet to catch strikes.

But the test is not correct as ND point out a good method from John Patricks baccarat book.
So you play 122 and it you catch a win with the first two then you continue with regression up and pull.

I also test Wells staking plan ... for example 456 and regression 4321...



There is two way playing pattern breaker.
One is to track the 8 patterns and wait until you get one sleeping and bet against it.
Then the odds is 7 to 1.

The other way is to track 8 patterns to repeat.
Once you get one repeat you wait for a new pattern and play, then you have the odds 7 to 1.
You play twice and betting against three patterns to repeat three times in a row.

[falkor]

Might have to try ego's variation, but 140 spins is not really enough to test a system like this that relies on waiting 60 spins.

Here's the next result for HLH:

Total wins: 255, Total losses: 38, Max wins in a row: 24, Max losses in a row: 2, Tally: 47, 37, 38, 42, 30, 36, 30, 33

Total won (without progression): 510
Total lost (without progression): 532

4/8 other patterns would have probably made a profit instead of betting against the missing one.

[falkor]

HLH - based not on the missing pattern - but on the previous pattern to appear:
Total wins: 280, Total losses: 36, Max wins in a row: 45, Max losses in a row: 3, Tally: 32, 47, 36, 36, 36, 38, 43, 48
Total won (without progression): 560
Total lost: (without progression) 504

Repeating that same test but over more spins.

[falkor]

This is the previous test repeated, but after 3,000 sets over the course of 1 million spins:

Total wins: 2624, Total losses: 340, Max wins in a row: 48, Max losses in a row: 3, Tally: 374, 363, 340, 398, 346, 375, 392, 376
Total won: 5248
Total lost: 4760

That seems pretty consistent and there was still no more than 3 losses in a row! 

Back to HHH for the same variation.

[ego]

-

I just show this, because you don't need marty to playing pattern breaker - regression up & pull works fine

[falkor]

ego, could you please explain in more detail how regression up & pull works regarding pattern breaker? I don't quite understand; could you give an example?

[falkor]

Does your graph show the tracking spins?

[falkor]

This one did better as time progressed, but was losing at the start:
Total wins: 2558, Total losses: 329, Max wins in a row: 50, Max losses in a row: 3, Tally: 329, 353, 365, 340, 377, 369, 376, 378
Total won: 5116
Total lost: 4606

So... based on the last pattern seems the way to go - not based on the missing pattern.

[falkor]

Going back to JL's rules on HHH has also resulted in more profit than loss, but only during the long-term:
Total wins: 1586, Total losses: 212, Max wins in a row: 36, Max losses in a row: 3, Tally: 212, 243, 220, 213, 205, 209, 259, 237
3172
2968

Next I will play all bets based on JL's rules (the missing pattern) and my variation (the last pattern to appear)

[falkor]

Using the original rules, JL's Pattern Breaker seems to fail in the long run. I'm at 800,000 spins, but need to move on to the next test.

Total wins: 14449, Total losses: 2091, Max wins in a row: 55, Max losses in a row: 4, Tally: 2038, 2037, 1950, 2100, 2073, 2097, 2115, 2130
won: 28898
Lost: 29274

Next tests will be:
1) my variation
2) betting on the missing pattern (not against it!)

[falkor]

My variation is already on a profit after 100,000.

My prediction is that most of the profit will come from betting on the missing pattern or betting against the most fluctuated pattern.

But you will need to remove the house edge from the final profits because I don't have the zero factored in this... BV style! 

[falkor]

Nope the losses caught up and the profits are only very slight, so would fail with zero.

Maximum losses was 5 in a row too.

Onto penultimate test...