Author Topic: The behaviour of variance, past spins and entry points  (Read 2324 times)

0 Members and 1 Guest are viewing this topic.


The behaviour of variance, past spins and entry points
« on: June 06, 2017, 05:31:02 PM »
According to probability, the average expectation after 101 spins is roughly 49 Blacks, 49 Reds and 3 zeros.
Variance is the the difference of the actual results from this theoretical expectation.
If in our game we get 60 Reds, 39 Blacks and 2 zeros then the variance is in favor of Red and against Black.

If we continue to play variance will change and the relation between Blacks and Reds will fluctuate.
Although variance (luck) is unpredictable it has some "rules" it follows.
First of all it can't be constantly in one direction. That is, it can not constantly favor one outcome, it has to change sides. This can be observed experimentally and it is based on the ergodic hypothesis.

However this knowledge, like other knowledge we have about the behavior of the wheel, like the law of thirds etc. is not easily exploitable.

In the following picture you see the behavior of variance over time.
What I said above could be interpreted as "what goes down must go up and vice versa". And this is true for the diagram below. However you'll notice this fact is not easily exploitable.

Both points A and B are after variance has gone up. After A variance keeps going up, while after B it corrects the imbalance very quickly.
Both points C and D are after variance has gone down. After C variance keeps going down, while after D it corrects the imbalance very quickly.
So yes, variance can't go one way forever, but it is not as easy as it may seem to define entry points that will guarantee a specific behavior.

« Last Edit: June 06, 2017, 06:14:41 PM by kav »
The following users thanked this post: Jake007, scepticus


Re: The behaviour of variance, past spins and enter points
« Reply #1 on: June 06, 2017, 05:47:27 PM »
Im going to have to think about this more. I would love to plug in the data and express it as a oscillating technical indicator.. something we would see on a stock chart. Because what happens with reversal points like B and D normally doesnt happen immediately. I would think its a gradual shift back to the median. Or??? maybe those reversal points ARE sharp. I dont personally know. Regardless, if the data could be expressed as a graph what inputs would be needed?

- spins
- red / black



Re: The behaviour of variance, past spins and enter points
« Reply #2 on: June 06, 2017, 06:05:21 PM »
hi Jake,

The reversal points could be both sharp or long winded as the diagram shows.
In order to get a diagram like the above I believe you need spins for X axis.
For the Y axis you need the moving results of a specific number of spins, say the latest 10 or 100 or whatever number of spins. So after each spin we have the same results and only replace the oldest result with the new one. Naturally different sizes of spins will produce different charts.

The following users thanked this post: Akanni


Re: The behaviour of variance, past spins and entry points
« Reply #3 on: June 06, 2017, 11:47:37 PM »

  hi, years ago when playing blackjack, which a player wins 47 of his hand and the the dealer 52, and pushes are if Iremember around 6%, I did a experiment that had very surprising results.
I am sure the results betting ec's should be similar.

every time I would be down 20 units within 100 hands, or spins, I would simply grab 10 units, flat betting to either win or lose and record the results.
so it is all done flat betting.
I ran about 50 sessions like that in my old computer.
the results were ;  70% of the time I won 10 units.
the other 30% I lost.

50 sessions may not be enough to engrave it on stone, but it is a nice result, considering risk/verses reward.

I also notice that the longer it took to win the 10 units, the more the forecast had become negative, which mean that if it takes too much time, you may want to abort the mission and take a small win or loss instead.

another thing is this; within 100 spins statistically you can expect to get 3 reversals. so if you get on a positive side and then go negative, and you do that twice, you could also take it as a signal of either trying a third time, which really end up being your last reversal into positive territory, or stop the session and call it a day, which I would rather do.
another point is to use a medium size bet if you play flat bet, so that if you happen to win 10 units early, you could regress to insure your profit.
only 2% of the time will a player win 20 units or more playing 100 spins. asking for more is way too much.

but I would play that method with pen and paper and make sure i know where i am at all time.
so multi level flat bet is necessary if you are going to wait for the - minus 20 units before kicking the 10 big ones to either win or lose. but remember you can stop at anytime when you are on the plus side and are only risking 10 big units. would only take one winning session to make it up.

The following users thanked this post: kav


Re: The behaviour of variance, past spins and entry points
« Reply #4 on: June 07, 2017, 08:00:40 AM »
Great post Rinard.
What do you mean by multi level flat bet? Betting 1 unit for 10 spins, then 2 units for another 10 etc.?
Wouldn't be more "safe" if instead of one big bet we did, say, 3 bets of 4 units? (like in Poker sometimes when people re all-in the run the hand 2 or 3 times to reduce variance)


Re: The behaviour of variance, past spins and entry points
« Reply #5 on: June 07, 2017, 08:00:59 AM »

 Kav if don't think this add to the discussion, then you can delete my post and not make a separate topic.

 Here is a experiment i done to observer variance with one extreme situation.
 I pick 10 events in a row using even money position and look at the next 10 results. 
 We know that 20 in a row show once in 1 million so there is plenty of observation to see how variance act.

 To get several samples with 10 in a row you can charting and tracking the streets on a rolling basis.
 Six streets (same as 18 numbers) hit around 7, 8, 9, 10, times in a row several times during 300 trail sample.
 So each time you get six streets hitting ten times in a row you just look and charting the next ten results.
 Then start over the tracking and charting searching for the next ten results in a row.

Another way is to look at the SD where 20 contra 2 is 3.83 SD.
That means 10 contra 1 is 1.92 SD
So you expectation would be getting at least two more hits after 1.92 SD if you would be betting against 3.83 SD
Then you would betting variance as regression.

Here is one example using lower values 12 contra 2 is 2.5 SD and pretty strong imbalance.
You can see this events as a window of events creating imbalance where some events are overrepresented.
So when you have 6 contra 1 you can pick the alternativ to follow the strength of the variance to grow stronger or you can bet for regression. Now if you would bet for regression you expectation would be at least 12 contra 3 or more in any combination after 6 contra 1. If you pick to follow the strength then you would assume not getting 3 contra 12 in any given combination after 6 contra 1.

So this open up for selection method with trigger and expectation betting with or against the variance.
3.0 SD window would be 14 contra 2 and the beginning could start from a window around 7 contra 1 in any combination.


« Last Edit: June 07, 2017, 08:04:30 AM by Sputnik »


Re: The behaviour of variance, past spins and entry points
« Reply #6 on: June 07, 2017, 08:08:18 AM »
Sputnik this is relevant but I lost you with "contra".
What does "14 contra 2", "7 contra 1" etc. mean? Can you please be more specific?


Re: The behaviour of variance, past spins and entry points
« Reply #7 on: June 07, 2017, 08:28:52 AM »

contra means you have 14 reds against 2 blacks

Well if you have 7 reds and 1 black in any combination.

Then if you bet variance will continue to grow stronger you will bet against black and follow red.
If you would bet for variance to get weaker (regression) then bet black.

But the selection method is based upon a window of events or some kind of determine imbalance.
Just to give you some kind of expectation.
So 14 reds and 2 blacks is 3.0 SD
Now if you bet for the 3.0 SD to happen or go in that direction to reach higher SD values you would bet red.
If you would assume the SD not reach 3.0 SD you would bet black.

But that is fuzzy so we can define what it means when variance act in this situation.
14 reds and 3 black or more and you have regression or variance make a drawdown peak.
14 reds or more and 2 black or less you have the variance growing stronger to reach 3.0 SD or higher.

So when you have 6 reds and 1 black and use a window of 2.5 SD as benchmark or reference or limit for your window of events to determine to play with or against.
Could be any and you can set a window for 3 , 4 , 5 SD

For example 10 reds - if you get 3 blacks for the among the next 10 reds or more blacks you will have a lower value then 3.83 SD and if you get more then 10 reds and less then 3 blacks you will have more then 3.83 SD

The point is that in some situation you know that if you get 3 blacks you will have regression and if you get less then 3 blacks in that particular situation you will have stronger imbalance.
This work if you set a benchmark to work with or a certain window of event.

For example assume you say that you will get windows of regression more then does window will reach 3.0 SD and you start betting from 7 reds and 1 black situations. Then you have create a situation where you have some degree of expectation based upon windows of 3.0 SD. You know that each time you get 3 blacks you have regression so you don't need to do anything before you get two blacks to show.

Or the other way around and you assume you will get imbalance all the time.
Then after 7 reds and 1 black you will not for the next 8 trials get a total of 3 blacks.

« Last Edit: June 07, 2017, 08:36:49 AM by Sputnik »
The following users thanked this post: kav


Re: The behaviour of variance, past spins and entry points
« Reply #8 on: June 07, 2017, 09:46:09 PM »

  Kav, what I mean with multi level is just that' ;  when you first enter a game you dont always know if you are going to bet black or red because you have no history.  so you can play a "I dont know bet" with a medium size bet", lets said a 20 $ bet, see which direction it is going. if you win 10 units early, you got 200$, and you can come down to a 5 $ bet if you wish to continued playing.
now if you dont win and lose 20 units, (400$), then you can pull out your 10 big bet of 100$ and go for a 10 unit win, which will get you 600 $ net profit if you win it.
now one can be more conservative and bet lower units of course. but I did experience many reversals when I tested sessions like that.
also if you wish to cover the 0/00, that can help. but if you dont and see a few zeros come out, please use it as a signal to go for a lesser amount of units. substracting the amounts from your 10 unit win.
so you really have 3 different bets; medium, low, and high.
50% of the time you win 10 medium size bets of 20$
the other 50% you will lose those 10 bets.
for the rest of the time after the losses you can win 7 out of ten times your big bets. not too shaby i think
 a easy method really.


Re: The behaviour of variance, past spins and entry points
« Reply #9 on: June 07, 2017, 10:00:08 PM »
Rinard thanks for the rely.
I lost you when a 10 unit bet ($200) will give you $600 profit.
I like the idea of low, medium and high bets, but I'm not sure how you use them.
You start with medium and if it goes well you go to low and when medium goes bad to go to high?


Re: The behaviour of variance, past spins and entry points
« Reply #10 on: June 07, 2017, 11:17:34 PM »

kav, maybe I did not explain well.  winning the 10 big bets of 100 $ = 1000.   - 400 $(20x200$) = 600 $ profit.

so you come to the table, said you lose 20 medium bets of 20 $. you are down 400 $.
since it is your trigger for the 10 big bets of 100 $, you go on and try to gain those 10 units of 100 $.  but since you lost the 20 medium bets (400$)  , then win 1000$, you are now up 600 $.
hope that clarifies it, R.