Gambler's Fallacy to 4 decimal places

[Colbster]

I had one of those dreams last night just as I was losing consciousness that has given me some adequate results in a little bit of hand testing.  Using the no-zero BV table, I am tracking free spins on auto-spin (2 units each on Doz 1 and Doz 2, 1 unit each on 25-30 and 31-36 is a free spin if you don't have one that you prefer).  In an Excel window, I list the results as they come up, usually taking just about a minute or so to get my tracking spins.  In the next column over, beginning after the 20th spin, I run an average function (the formula for non-Excel users is =average(A$1:A20) in cell B20 if you are tracking top to bottom in column A.  You would then copy this formula into B21 down through B40, giving you a total of 40 spins and 20 averages).  The average of all spins in the long term will be equal to 18.5 (666/36=18.5 or consider the average of 1 and 36, 2 and 35, 3 and 34, etc will all be 18.5).

After we get our first average following our 20 spins, we compare our average to 18.5.  If it is below 18.5, the law of averages says that we will get more "high" numbers to bring the average up.  If it is above 18.5, we would expect more "low" numbers.  Why this is slightly different than just keeping a running +/- count is that we have a more exact idea of how the numbers have skewed, as well as a better idea of the degree.  If we are just a little away from 18.5, for example 16.7, I would play 1 unit on "high" for the next 20 spins.  Since it is an EC flat bet, there are very low risks involved.  However, if they variation from the average of 18.5 is more significant (I use +/-3 as my line), I move my bets up to 2 units on high or low for the next 20 spins, meaning 2 on High if we have 15.5 or below, 2 on Low if we have 21.5 or above.

For the statistics guys, since we have a small sample, the size of the standard deviation is fairly large, with 1 standard deviation being closer to 13.5 and 23.5, but this works nicely for the short term and gives us more betting opportunities.

Note that we will continue tracking each spin in our spreadsheet as it hits, and we need to pay close attention to the averages as they continue to be calculated.  If I started at 15.5 with two units on high, but get several quick 33s or 36s, I am glad to have won some money but the average will have quickly approached 18.5.  If it gets between 17.5-19.5, I stop the spins and re-track.  If I happen to have my first 20 fall within this range, I don't bet.  It is easy enough to just get a new set of 20 spins and I don't see that there is a real statistical advantage with so few spins comprising our data set.

Just initially, I am using 40 units as my bankroll in a session.  My max drawdown so far has been 3 units in a 20-spin set, but I have never ended worse than 0 after 20.  Most won so far is 5 units in a shortened 7-spin set (I hit my 18.5 range and stopped while I was ahead).

[chrisbis]

@Colbster.


Nice idea mate. I like it, and got it straight away.


Question for U:- How long(deep) have U tracked for, after Ur initial 20 spins?
.....and then hence, how long before U start over again and 're-track'? (average answer!)


cheers Chris


P.S. a screen shot, or actual Xxl file of Ur excel sheet would be great to see posted.

[chrisbis]

I just had a real wacky thought.


But.............it could just turn out to be reasonable one in the end.


What would happen, if we ran Ur Averages on Hi and Low, directly against one of the JL methods?


That being say Averages V Pattern Breaker.
It would show, if the Bet that is exposed in the PB, matches up with the expectation of where the average sits against Ur Model 18.5


By having this Dual approach, when the indicators are very strong, and alignment is confimed, then the bet could be raised even higher.


I imagined, have the two sets of systems laid out side by side, and a little program running of when the expectation was at its highest.


For example, if U had a comparison of a running average of 13.5 and the Pattern Breaker,(or any other EC Hi/Low method for that matter) was indicating to bet high also at that moment, then U could raise Ur hand for a High bet by a factor of x5


(assuming U where running x2 bets base as U show/mention in Ur explanation, for a +3.5 smaller/greater than 18.5....then the bet would then be a base bet x 10) (2x5=10)


A whole array of smaller increments against Ur normal/standard bet could be formulated.


Say if 17.5-19.5 on Ur system, Colbster, U would not bet, but if one of the JL series Matrix was showing a bet, then U would just use 1 x base bet value.


If 17-20 was averaging, + One of the EC's was showing a likely bet on say PB, then we could have a x2 factor on our base bet.


And so on.
Range set the multiplier.


16-21 then x 2.5
15-22 then x 3
14-23 then x 4
13-24 then x 5 (which could be doubled, since U would be doubling up with Ur 4 decimal places )
12-25 then x 7 (doubled up = x 14)


Just a thought as I said at the Top of the story!
cheers.

[Colbster]

I love your idea of quantifying how much to bet on another strong system's bet selection method.  It would be a piece of cake to extend the spreadsheet beyond the 40 bets that I set it up with to include the longer tracking period of one of JLs methods.  Obviously, this would only be useful for tracking H/L and would not help with B/R or E/O, but it would not be that hard to create an integrated Excel sheet that did both on the same grid.  In fact, I will start work on that immediately and post when through.  I'm not coder, but I can make my way around a spreadsheet fairly well.

In replay to your question about tracking depth, I have never tracked beyond 20.  I have been at break even or better by the time I have gotten to that point every time (although in early stages right now).  I just keep a mental tabulation of how many I am up or down, which is easy since we are only +1 or +2 and -1 or -2 at any point.  If I was down 1 or 2 after 20 spins, I would probably continue to break even if the average still looked favorable to continue, but I haven't gotten hard rules laid down for that yet, which is why this is in Notepad instead of Full Systems.

Thanks for the feedback.

[chrisbis]

Hi Cobber.


If You can take an average for Hi/Low, then why can you not take the same average for Red/Black and
correspondingly, for Odd/Even. I can not see why the average 'Shift' is not the same, its just distribution from a shared Centre point.


I think, (in answer the your Pm on the Matrix) you should open that question out to the Matrix Users, over in one of the JL systems, or open a new Topic, just to pose this question.


Q. Which matrix is best suited to................................?


Cheers Cob


P.S. This "Qualification" process could extend into areas neither of Us have yet to think of!


I think EVERYONE on the forum, who plays Roulette with any of the Matrix systems, should come to this place, and re-read your opening idea, and then my following 'bolt-on' theme, and add to our debate.


This could really go somewhere!

[Colbster]

It isn't really an average of Low/High, more an average of the actual spins.  With a strong enough bias away from the center, we might even be able to tighten it up to 1st or 3rd dozen, instead of just High/Low.

For the ECs (H/L, E/O, or B/R), we would just track and divide 1 number by the other.  For example, 9 B and 11 R would be 9/11 or 0.818181.  That would be the equivalent of 15.136 as an average (0.8181818*18.5).  That would be more work than would be necessary, though.  If we had 8 or fewer of an EC in 20 spins, we would bet for than side of the EC for the next 20, anticipating that it would come back closer to the average of 10.

[superman]

If I am reading right you are really talking about standard deviation, a fixed 20 spins isn't really going to help you find anything, the overall picture would cover probably up to 50 spins, and within those spins your trigger measurement would look something like this

In 15 spins (2 reds / 13 blacks) = 13.33%
In 50 spins (14 reds / 36 blacks) = 38.8%

But you may be suprised just how long the bias can last even after a trigger has formed, a 15 spin trigger can take you right up to the 50 spin area, and if it does any progression you use would probably be costing a lot at that point. Just like runs and chops, the bias we know must change but it changes at a different time every time. Good luck.

[ZeroBlue]

But you may be suprised just how long the bias can last even after a trigger has formed, a 15 spin trigger can take you right up to the 50 spin area, and if it does any progression you use would probably be costing a lot at that point. Just like runs and chops, the bias we know must change but it changes at a different time every time. Good luck.

i tend to disagree.
From my experience, the shorter the window and the larger STD the better.