Our members are dedicated to PASSION and PURPOSE without drama!

Return to the Mean

Started by albertojonas, April 01, 2015, 11:18:26 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

albertojonas

Hello all.

Suppose you get 10 spins of an even chance, and compare them with the next ten spins. You will get at both extremes either 10 same or 10 different.
I test this with random.org and i will put here some examples so you can make your own observations and conclusions.

Timestamp: 2015-04-01 23:11:01 UTC
spins [823,852]

2 2 2 2 1 2 1 1 1 2
2 2 2 2 1 2 1 1 1 2
1 2 1 1 2 2 2 1 1 2

WLWWWLWLLL

[Edited to explain better]
Cheers

Tomla

so basically anything can happen ,,, either the dominant will slowly fade or fade very quickly---the devil is in knowing when

but the over dominant in the second set was much stronger in the first 10

albertojonas

Thing is that after a very strong imbalance you can expect correction to happen soon. And that means that you will win as the very strong imbalance tends to fade. This illustrates RTM.
There is no need to wait for a rare series as any set of 10 as the same probability.
If you can devise a betting algorythm to take advantage you will win most of the time. I personally flat bet.

yet another example
from yesterday's randomization

2 1 1 2 1 1 2 1 2 2
1 2 2 1 2 2 1 2 1 1 -second set is exact opposite of 1st set so there is huge std towards FTL bet
2 1 1 1 2 2 2 1 1 2 - we will bet the 3rd set will match 1st set

WWWLLLWWLW

One more

1221212222
2112121111
2121122122
LLWWLLWLWW

______________________________-

This is for illustration. One can compare OLD or FTL selections in different lenghts and for several ec's.

albertojonas

Correct me if i am wrong, but the above scenarios have the same probability of ten blacks followed by ten reds or twenty reds or twenty blacks...

Tomla

I would imagine yes--have you been noticing something alberto

Sputnik

I never forget this quote from WIKI:

"Regression toward the mean simply says that, following an extreme random event, the next random event is likely to be less extreme. In no sense does the future event "compensate for" or "even out" the previous event, though this is assumed in the gambler's fallacy (and variant law of averages). Similarly, the law of large numbers states that in the long term, the average will tend towards the expected value, but makes no statement about individual trials. For example, following a run of 10 heads on a flip of a fair coin (a rare, extreme event), regression to the mean states that the next run of heads will likely be less than 10, while the law of large numbers states that in the long term, this event will likely average out, and the average fraction of heads will tend to 1/2. By contrast, the gambler's fallacy incorrectly assumes that the coin is now "due" for a run of tails, to balance out."

So i run 1 million trails to see the worst and extreme for even money bets:

Low: 18 in a row (once)
High: 18 in a row (once)
Red: 19 in a row (once)
Black: 20 in a row (twice)
Odd: 18 in a row (twice)
Even: 18 in a row (once)

Bayes

Quote from: albertojonas on April 02, 2015, 12:06:45 AM
Correct me if i am wrong, but the above scenarios have the same probability of ten blacks followed by ten reds or twenty reds or twenty blacks...

Hi aj, nice to see you posting again. You are correct, there's no difference so no point in waiting for ten blacks/reds in a row (or any other trigger for that matter). I use a similar kind of moving template/pattern scheme myself, but probably in a slightly different way. I always revert to the start of the pattern after any win, I don't play it through to the end regardless.

Suppose my current pattern to bet against is R BBB RR B R (I generally pick a pattern between 8-10 spins long).

Here's a short example of how I adapt betting as the pattern unfolds.

1. Bet against R (element 1 of the pattern), so continue betting B until a R occurs.
2. R hits. Now I bet R because I'm betting against element 2 (B). So I just continue to bet R until a loss, which means a B has hit.
3. When B hits, I revert to step 1, so now betting B again.

So for a lot of the time, I'm betting FTL, it's only after losses that I'm actually betting against the pattern. I find that this keeps losses manageable and you get some great winning runs.

Of course, it can happen that the first part of the pattern repeats, in which case you will get quite a few more losses than wins, but a suitable progression will neutralize them, or you can just flat bet and make up the losses later when the winning runs come, and they always do. How I actually pick the current pattern is complex, but just using random patterns works pretty good too.




AsymBacGuy

Imo, Sputnik and Bayes provided very good replies.

I'd add the value of the decline in probability concept thanks to a past contribute of sxzbox (or something like that) member.

In a word and for the few who don't know the concept, rarer events tend to appear clustered at the start then their appearance will dilute.
Despite the original work (Spencer-Brown) didn't suggest a possible gambling advantage, I found that many rare occasions tend to be clustered, then deeply decreasing their frequency.

Since we cannot establish a "start" to many rare gambling situations, we should wait the times when they will appear very clustered, then betting toward the opposite events' apparition up to some point.

Curiously, at least at baccarat, some unfavourable rare situations will tend to follow such distribution, meaning that our "enemy" will come out very often or almost nothing at all.

as.   







 

Baccarat is 99% skill and 1% luck

CLEAR EYES, FULL HEARTS. CAN'T LOSE
(Friday Night Lights TV series)

I NEVER LOSE.
I EITHER WIN OR LEARN
(Nelson Mandela)

Winners don't do different things, they do things differently (Albalaha)

Sputnik


I test this and got the code:

10 trails is 4.52 SD (20 in a row
11 trail is 4.74 SD ( 22 in a row
12 trails is 4.96 SD ( 24 in a row
13 trails is 5.16 SD ( 26 in a row
14 trails is 5.35 SD ( 28 in a row
15 trails is 5.54 SD  ( 30 in a row

This is when you take for example 10 random outcomes to match 10 future ones.

Is not working,
I run 2000 trails several times and when i up 20.000 trail almost all have hit 20, 22, 24, 26, 28

So this is not better then betting agains't 10 11 12 13 14 15 in a row

albertojonas

when you take for example 10 random outcomes to match 10 future ones, what you are getting is a string of wins or losses. On both extremes LLLLLLLLLL or WWWWWWWWW. So what we are really measuring is FTL bet result.
The example for ten random outcomes is 3,16std.

observing 10.000 outcomes, comparing 14 vs 14 on a rolling basis, these are the number of wins i got:

0   2
1   13
2   79
3   260
4   632
5   1214
6   1851
7   2081
8   1726
9   1169
10   633
11   234
12   69
13   10
14   0

albertojonas

using ftl to match last 10 outcomes, one can get a string of WL results. Those results act as an additional even chance outcome. Following this logic we can use 11vs11, 12 vs 12, etc... Generating as many EC possible bets as we want, shortening the periods of tracking.
This is particularly usefull for those who seek extreme events, std, marigny, rtm principles.
Cheers

Tomla

Albertojonas---most say to play against the last 8 or 10 decisions---are you saying to replay the last 10?  i have tested going against and it doesn't do well in my testing : )

albertojonas

Say u always compare last ten with next ten decisions.
After that u measure std or ecart (as marigny) and you decide to start play at 3.5 std.

I test this in random.org

13/04/2015
L(LL)(wwwww)+2
W +1
W +1
W +1
LL -2
WWWWWW +6
L -1
WWWW +4
----------------
12/04/2015
L
LL
LLLL
LLLLL
WWWWWWW
WWWW
W
(W)
----------------
11/04/2015
L
WW
L
WW
W
W
L
WW
W
L
WW
W

Drazen

My first test test was pretty interesting:

2 2 2 2 2 2 2 2 1 - first ten decisions
1 2 2 2 2 1 2 2 1 - second - start to bet against first formation
2 1 2 2 2 2 2 1 2- oh dear we run on pretty nasty storm here; - L W L L L L L L W W; continuing to bet against first formation
1 1 1 1 2 1 1 1 2 - ka boom; and the sun has shined hehe
---------------------------------------------------------------------------
+1 flat bet overall

Cheers
Common sense has become so rare it should be classified as a superpower.

Sputnik


@albertojonas

I don't understand how you play FTL using RTM - can you explain step by step.

I understand how Drazen does with his example.

Now i will show you have i experiment, for example if i take 8 contra 8 (random events) then if i get 14 same and 2 oppisite i have 3.29 SD.
This means that if i only get same i am not betting as there is no RTM present, but after i have two oppisite to show i might bet.
That way i play agains't window of propability of 3.29 SD

Now i can change my benchmark to 4.08 SD window of probability that would be 11 contra 11

This is how it looks like: WWWLWWLLWWWLW


2
2
1
1
2
1
1
1

2 S
2 S
1 S
1 S
2 S
2 O
2 O
2 O W

2 S
2 S
2 O
1 S
1 O
1 O W
2 S
1 O

1 O
2 S
2 S
2 O
2 O W
2 O
1 O
2 O

1 S
1 O
1 O
2 S L
1 O W
2 S
2 O
2 S

2 O
1 S
1 S
2 S
1 S
1 O
1 O W
1 O

1 O
1 S
1 S
2 S
2 O
1 S L
1 S L
2 O W

1 S
2 O
1 S
2 S
1 O
2 O W
1 S
2 S

2 O
1 O
2 O W
1 O
2 O
2 S
2 O
2 S

1 O
1 S
1 O
1 S L
1 O W.
1 O
1 O
1 O