Exploiting the unusual - the sequel

[Arteinvivo]

On G.G, I wrote in the Bac section that I have some inclination to develop approches that try to profit from periods of stability where two aspects of a proposition get a bit blurry as if it was not clear what was the tendency of the moment. Is it producing streaks or chops ? I qualify these times as island of stability or neutral periods. This is what my system tries to exploit.

When a new idea pop up in my mind, the first thing I look at, is its nemesis. Your chance to stay in business longer is dependent upon the nemesis of your particular idea or plan. If the nemesis is too evident to find then the system is generally not good at delivering a steady profit. Again, this aspect is terribly tough to get right using a mechanical approach, the human mind is more apt to cope with this fuzzy logic (uncertainty) then a computer. But it does not prevent me to try by developping simple rules.

This is in this spirit I tried to develop this new idea which revolves around the tracking of "singles vs series".

THE SYSTEM

Here is a small sequence with a mix of streaks and singles :

R
R
R
     B    1 up < increase count +1 when you observe a streak
R        0 dw < decrease count -1 when you observe a single
R
     B    1 up < increase count +1 when you observe a streak
R        0 dw < decrease count -1 when you observe a single
     B   -1 dw < decrease count -1 when you observe a single
     B
R        0 up

Here is my suggestion to track series / singles during the course of a session. Keep a running count which will increase +1 when you observe a streak and decrease -1 when you observe a single. Look at the third column, it contains this running count (1,0,1,0,-1,0).

Let's look at the running count using a graph :


Look at the three horizontal lines (1,0,-1) representing the running count. We can easily see the horizontal line, which crosses the 0 on the vertical axe, holds 3 dots meaning our running count passed 3 times at this level.

This is precisely what the system wants to exploit, the dominant horizontal lines which form during short sequences.

I have looked at it and tested it from different angles and I came up with the following set of rules in order to play and target the dominant line of the moment in the most efficient way I could find :

FINDING THE DOMINANT HORIZONTAL LINE
Again, I'll repeat it differently as this is at the root of this system. The horizontal lines represent the value of the running count at specific intervals in time. Each dot represents an occurence or the passage of the running count at that level. In order to find which horizontal line is the most dominant at any one time, we need to log the frequency of these occurences or passages for each line individually.

For example, in the first table of this comment we have had these values for the running count : (1,0,1,0,-1,0).

You can look at it this way :

count : frequency
-----------------
   -1 : *
    0 : ***
    1 : **
[/size]

[Arteinvivo]

[size=12]
RULE #1
When a series or single is clearly identified, you update both the running count and its associated frequency. You look at the two highest frequency values in your log. When there is a difference of 1 between both then you select the count which shows the highest frequency.

Ex. :

R
R
R
     B    1 up
R        0 dw
R
     B    1 up

At this point, we have this log :

count : frequency
-----------------
    0 : *
    1 : **

Looking at the two highest frequency values in this log (not difficult as we have only 2 values at this point) we note a difference of 1 between both, so we select the dominant count 1 as it shows the highest frequency.

count : frequency
-----------------
    0 : *
    1 : ** < dominant count identified : 1

In short, we need the frequency of at least 2 count values and a difference of at least 1 between the two highest frequencies. When there is equality between those 2 highest frequencies we do nothing.

Let's suppose we have this scenario where 3 count values (-1,0,1) are logged and with already a leading count identified :

count : frequency
-----------------
   -1 : *
    0 : ** < dominant count
    1 : *

The count value 0 is the leader or dominant count but when a new series or single is detected we get this new updated log :

count : frequency
-----------------
   -1 : **
    0 : ** < dominant count
    1 : *

Since 0 was already the current leader, we don't change its dominant status. In fact, when a dominant count is selected, a replacement will occur only if a greater frequency of +1 is found. The equality shown here is not an indication strong enough to conclude a new leader is emerging so in those cases we keep the current dominant count active.

[Arteinvivo]

[size=12]
RULE #2
Once we have identified our leader, we start to bet only when the running count moves one position away from our dominant count. Ex. if the dominant count is 1 and our running count becomes 0 then we are one position away from the dominant position. The same is true, if the running count becomes 2 then we are one position away from the dominant count or position. When this occurs, we suppose the running count will have a tendency to return to the 0 level as it is the current dominant position so we bet on Red/Black accordingly.

Using the same sequence :

R
R
R
     B    1 up
R        0 dw
R
     B    1 up < dominant count identified : 1
R        0 dw < one position away from 1

At this point, to see the running count returns to 1, what color do we need to back ?
Red as to go from 0 to 1 the count needs to increase and as mentionned in the tracking paragraph, the count increases +1 when we observe a streak. A streak at this point means Red would keep hitting.

So what happened ?

R
R
R
     B    1 up
R        0 dw
R
     B    1 up < dominant count identified : 1
R        0 dw < one position away from 1
     B   -1 dw < we lost

When we lose, we artificially set the frequency of the current dominant count value equal to 1 and search a new dominant among the frequencies of our count logged up to this point. In this example, the dominant count 1 had this frequency 2 so we change it to 1 in order to identify a new leader among our new updated log :

BEFORE UPDATE:

count : frequency
-----------------
   -1 : *
    0 : **
    1 : ** < dominant count

AFTER UPDATE:

count : frequency
-----------------
   -1 : *
    0 : ** < new dominant count
    1 : *  < previous dominant reset to 1

OK now we have a new dominant count identified which is 0 and we are at a spin position which coincides with the detection of a single which place the running count at -1 at exactly one position away of our dominant count.

So to see the running count -1 returns to 0 what color do we need to back ?

Here is what we have :

R
R
R
     B    1 up
R        0 dw
R
     B    1 up
R        0 dw < dominant count identified : 0
     B   -1 dw < one position away from 0
? / ?         < what color do we need to back ?

Remember, we increase +1 when we observe a streak and decrease -1 when we observe a single. Then to see the running count returns to 0 we'd need to see a streak. The answer is Black as we need to bet Black to see a possible streak of Black at this position.

And fortunately this is what we got.
[/size]

[Arteinvivo]

[size=12]Here is a more elaborate example using spins from Wiesbaden table 2 - 25 February, 2008.



Which dominant line do you think has brought the final profit of at least +1 ?

A cue, look at the line which crosses -8 on the vertical axe

I am using a simple Alembert progression where i increase +1 after a lost and decrease -1 after a win.



[/size]

[Arteinvivo]

[size=12]When zero hits i don't increase my bet size :



Here is the sequence of spins which produced this graph :



[/size]

[Arteinvivo]

For those of you familiar with the procedure of betting the same as previous decision you will recognize this procedure which ressembles the dominance theory where your bets follow the current pattern of decisions. To accomplish this you need only make sure that each time you have identified a dominant running count that each of your bets will back what it needs to back in order to target the same dominant running count as the second preceding count.

If the current dominant count is -7 and you have this string of counts :

-7, -8, -7, -6, -7 then you follow the dominant count -7 by betting on the color which will bring the count at -7 every two series/single.

Hey ! after all, we don't need to reinvent the wheel each time.

[Arteinvivo]

[size=12]Wiesbaden stream today table 2 - February 26, 2008.

This begins to make a lot of zero, no ? but again the profit is there.  [smiley=vrolijk_26.gif]

Representation of the running count using a simple graph :



The datastream :

[/size]

[Arteinvivo]

[size=12]

Let's look at table 9 - Wiesbaden February 26, 2009



Data stream :


[/size]

[Arteinvivo]

[size=12]Out of curiosity, i changed my progression. Doubling after a winning bet such as this 1,2,4,8,16 never going higher than 16. I took 10000 spins from Hamburg and never lost a sequence. Conclusion, with this system, winning bets come in clusters.[/size]

[JonInRI]

Hey Arte, Another great system  

I was wondering a couple of things.  When you said you tested against Hamburg:

I changed my progression. Doubling after a winning bet such as this 1,2,4,8,16 never going higher than 16. I took 10000 spins from Hamburg and never lost a sequence.

I was looking for a little clarification on the "never going higher than 16" part.  Did you mean that the systematic progression you decided to use was to never go above 16 or did you mean in the 10,000 spins it never went above 16 but you would have been willing to if need be?

Also, I was wondering what you programmed this system in to test it.  The pics you posted don't look like RX, although I will admit I haven't been using it long [smiley=embarassed.gif].  It almost looks like you may have done it in Excel or some other spreadsheet program.

Thanks,
Adam

[Arteinvivo]

Did you mean that the systematic progression you decided to use was to never go above 16 or did you mean in the 10,000 spins it never went above 16 but you would have been willing to if need be?

Second part of your question is correct. I just wanted to know if the majority of my attacks would be won without ever going over 16. Would i end positive or not ? The answer was i ended positive.

But nevertheless there are better approaches while still using the same concept. I now use this concept on Red/Black without tracking series or singles but every decision. When Red shows i decrease my count -1 and when Black hit i increase +1. This is much faster and it is also easier to follow the current trend.

All simulations are done in Excel. The best tool in the west.

[dennisbelle]

 I now use this concept on Red/Black without tracking series or singles but every decision. When Red shows I decrease my count -1 and when Black hit I increase +1. This is much faster and it is also easier to follow the current trend.

Arte, could you give an example of the above?

[Arteinvivo]

Dennis,

If we take as an example table 2 - Wiesbaden 2008/03/05 (yyyy/mm/dd)

The first 36 spins excluding the zero.

We get this string of numbers :
-1, 0, 1, 0, -1, -2, -3, -2, -1, -2, -3, -2, -3, -2, -3, -4, -5, -6, -5, -4, -5, -6, -7, -8, -9, -8, -7, -8, -7, -8, -9, -10, -11, -12, -11, -10

and this graph :


The idea is to follow the current trend but by trend i mean everything which repeat.

[dennisbelle]

Arte,  would you explain/show how you would bet these numbers? [smiley=rolleyes.gif]

[Arteinvivo]

Dennis,

If we take as an example table 2 - Wiesbaden 2008/03/05 (yyyy/mm/dd)

The first 36 spins excluding the zero.

We get this string of numbers :
-1, 0, 1, 0, -1, -2, -3, -2, -1, -2, -3, -2, -3, -2, -3, -4, -5, -6, -5, -4, -5, -6, -7, -8, -9, -8, -7, -8, -7, -8, -9, -10, -11, -12, -11, -10

and this graph :


The idea is to follow the current trend but by trend i mean everything which repeat.

[size=12]Dennis,

Remember, we are playing on an even chance proposition. Here we have a string of Red versus Black. The way i track is as simple as keeping a running count. I would write it down to avoid errors. So here is how i would have proceeded with this particular stream :

First, let's forget about Red or Black as this is irrelevant. We want to use our running count as our sole indicator.

Let's look at the running count as it occured :

-1
0
1 <- highest point
0
-1 <- i bet the count will return to -1 (Got it)
-2 OK i won last decision so i resume tracking
-3 <- lowest point
-2
-1 <- i bet the count will return to -1 (Got it)
-2 OK i won last decision so i resume tracking
-3
-2 <- i bet the same as last preceding count (Got it)
-3 <- i bet the same as last preceding count (Got it)
-2 <- i bet the same as last preceding count (Got it)
-3 <- i bet the same as last preceding count (Got it)
-4 <- i bet the same as last preceding count (*WRONG*)
-5
-6
-5
-4 <- i bet the count will return to -4 (Got it)
-5
-6 <- i bet the count will return to -6 (Got it)
-7
-8
-9
-8
-7 <- i bet the count will return to -7 (Got it)
-8
-7 <- i bet the same as last preceding count (Got it)
-8 <- i bet the same as last preceding count (Got it)
-9 <- i bet the same as last preceding count (*WRONG*)
-10
-11
-12
-11
-10 <- i bet the count will return to -10 (Got it)

So in all, we have : +12 units, -2 units.

As you can see making these educated guesses were not that difficult. This is not always the case but i can tell you long losing series using this approach is a thing of the past. Each sequence must be dealt differently. I never bet when there is a streak going on, i let it end its course then i make an educated guess. Losing 4 times consecutively is tougher using this approach than all other bet selection i have tried before.[/size]

[Arteinvivo]

-