Is it relevant at all? We know that regression analysis is widely used for prediction and forecasting, but we also know that application of this in small effects or cause-effect based on observational data, which is what practical application of this in roulette, can give misleading results. Then are we not setting ourselves for failure by mapping these principles into roulette? Strange thought, but I thought very relevant for a casual gambler and math guyz alike.

Quote from: Leapyfrog on September 03, 2014, 02:16:10 PMWe know that regression analysis is widely used for prediction and forecasting

There is something that intrigues me at least a little: If there may be a tendency of regression to the mean in a few sets as something accepted.

Many times after a series of red together it is sensed a little more black is coming. And often so goes black more than it should after that pile of red. And I put red and black as an example. It may well be something else.

This I see it in one sitting. I have no statistics on this phenomenon but rather as an observation. I wonder if maybe one should tolerably go to the casino without system and see if something catches this phenomenon. Then try to get in the plus.

Go see if there is a spontaneous streak to try to make a few units when it begins to go to the other side of the streak. Not millions of sets I clarify. but something that appears spontaneously within a normal play session.

Maybe someone has this figured already? How to take advantage of the return in the short term? and finally: is not it easier to be in the return zone if they are only a few deviant spins? Or is that these short returns are not accepted: only valid returns after millions of sets.

Thank you.

The gambler who expects to see "average" results grossly underestimates how large a sample size is needed for convergence to the average. The gambler's fallacy arises from the mistaken belief that mean reversion happens much quicker than would be predicted from the law of large numbers.

Quote from: Albalaha on September 04, 2014, 06:55:37 AM
The gambler's fallacy arises from the mistaken belief that mean reversion happens much quicker than would be predicted from the law of large numbers.

Absolutely. People expect that the characteristics of a short sample of outcomes will match those of a larger sample, but it's not the case. The gambler's fallacy is a very very stubborn cognitive bias, one which I'm sure the casinos are grateful for.

Quote from: Slacker on September 04, 2014, 07:56:00 AM
People expect that the characteristics of a short sample of outcomes will match those of a larger sample, but it's not the case.

Exactly and this goes back to my question on are we
[attachimg=1]
with roulette and statistical principles?
If this is true, then what is the difference between someone who says i am applying statistical principles and probability rules to play my game and some one playing a game of play the opposite? Does it matter really? What is the significance of variance and standard deviation if all it can do is analyse what has happened in the past? What is the use of bionomial distribution and its application in roulette if it can apply only on very large sets of sample and not on your session sample. It all sounds like a bit hyped up to me now.

Hi Leapyfrog,

This is an interesting topic and one which gets to the heart of what probability means. The idea that probability is a relative frequency is only one interpretation; others are more useful, depending on your needs.

You might be interested to read these two Wikipedia articles:

Frequentist Probability
Bayesian Probability

Regarding the binomial distribution and others, samples don't need to be very large in order for them to reasonably approximate the "long term" convergence value, but the smaller the probability, the bigger sample you'll need.

Quote from: Slacker on September 04, 2014, 11:49:53 AM
The idea that probability is a relative frequency is only one interpretation; others are more useful, depending on your needs.
You might be interested to read these two Wikipedia articles:

Slacker - Indeed they are and very informative. After reading through those, got quite a bit cleared and clarified. And you are right about bionomial distribution as well. Excellent and thanks.

Just started experimenting now. A simple thing in the lines of what am doing now.

In a sample of 5, ideal positioning of a Red to come is 2 times considering a single zero. Now as i understand expanding this sets of 5 further into sets of 30. Ideal positioning of a red to come is 12 out of 30 times, so on and so forth as am taking certain worst case consideration. However we do also have variance. In an ideal sample, 1 out of 4 sets of should have Red appearing less than 2 times. Extrapolating this 5 out of 20 sets/100 spins should have either 1 or no red.

How do i put all these together? I am currently trying out a simple method.
Pick any even chance Red or black or low or high or whatever you may choose.

Play in sets of 5.
1. Play 1 unit in first spin.
2. Win or lose play 2 units in second spin. If both 1 and 2 are wins, then set is finished. See through the next 3 bets without any bet or minimal bets.
3. If the one or both above is a failure play the third spin with 1 unit. If you have reached two wins on your selection by 3rd spin, end of set, else proceed to the fourth spin.
4. If all of the above were failures, place 5 units. On a win proceed to the 5th spin with 1 unit. On a loss end of set. If you had one win in the previous 3 spins, place 2 units. On a win end of set. On a loss place 5 units on the 5th spin.

I am choosing 3 even chances to play this game. Going slow and steady so far and am seeming to win more than i lose.

This is just a start and i think there are enormous possibilities and lots of reading to be done. The more i read it is getting more interesting. I am glad i started this topic. Following is a recent game that i played.

I have pre-selected low, even and red to play.

24 - 1 unit of L, E and R
5 - 2 units on L, E and R
18 - 1 unit on L, E and R   - End of set. wait for the next 2 spins to pass.
9
24 - end of set with +4 units
-----------------
31 - 1 unit on L, E, R
31 - 2 units
29 - 1 unit
32 - 5 units
6 - 1 unit on E and R. No units on L. End of set -7 units
---------------------------------------
15 - 1 unit on L, E, R
6 - 2 units on L, E, R. End of set for L
7 - 1 unit on R and E.
8 - 2 units on R and E
6 - 5 units on R. End of set. End of set -4 units
----------------------------------------
36 - 1 unit on LER
17 - 2 units
31 - 1 unit
21 - 2 units. Red set finished.
8 - 5 units on LE. End of set +4 units
----------------------------------------
18 - 1 unit
10 - 2 units. Set finished on Low and even
29 - 1 unit on red
36 - 2 units on red. Sets finished. Let one spin pass.
4 - End of set +6 units
---------------------------------------
30 - 1 unit of LER
27 - 2 units. End of set of Red
18 - 1 unit on LE. End of set of Even
10 - 2 units on L. All sets over. Wait for one spin.
20 - End of set +3 units
---------------------------------------
36 - 1 unit on LER
10 - 2 units. End of set for Even
25 - 1 unit on LR. End of set of Red
10 - 2 unit on L. End of set of L
1 - No units. End of set +5 units
---------------------------------------




There were 7 sets that i played. 35 spins in total.
End result : +11 units.
Win on 5 sets and loss on 2 sets.

As you have seen i have fixed the Even chance. For a moment, think that i have selected a different set.

High, Odd, Black - +5 units
High, Even, Black - +24 units
High, Odd, Red - -14 units
High, even, Red - +5 units
Low, Odd, Black - +11 units
Low, Even, Black - +30 units
Low, Odd, Red - -8 units

As you can see out of 8 possible combinations, 2 are losers and 6 are winners. If I had played them as 8 different sessions I would have ended with +64 units. That i think is the power of thinking this way. I am not saying this will win for ever. What am saying is a thank you to Slacker for shaping my thoughts in a different direction and I believe that is the right direction if my choice is away from the physics of roulette and treating them as a random set of numbers.

Well done Leapyfrog - what a fascinating concept! I guess you were a little bit lucky with this group of numbers - no zero. On average I would anticipate at least one zero in the mix for a 35 number selection. Have you got an example with a zero in it? And how do you handle it? Just take the loss and ignore it in the sequence? I think that is what I would most likely do.

Quote from: sqzbox on September 05, 2014, 12:19:09 PM
I guess you were a little bit lucky with this group of numbers - no zero. On average I would anticipate at least one zero in the mix for a 35 number selection. Have you got an example with a zero in it? And how do you handle it? Just take the loss and ignore it in the sequence? I think that is what I would most likely do.

You are right. I have been lucky this group of numbers with no zero.

With regards to your question on how to handle zero, i haven't thought about your way of ignoring in the sequence. Because am expecting 2 out of 5 number to be the even chance, i have provisioned for zero to come there. So I do take it as a loss, but not ignore it. See the following example going with my fixed selection of LER

17 - 1 unit on LER
0 - 2 units on LER
21 - 1 unit on LER.
10 - 2 units on LR and 5 units on E. End of set for L
11 - 5 units on R and 1 unit on E. End of set -7 units

As I mentioned, this is one element and not the only element. I am continuing my experiment and trying to mix several elements here based on this concepts like how to expand this as the sets and losses get larger and larger and how to vary how much you spend based on where it fits within in a bigger picture. It might turn out to be a futile exercise, but a very enriching experience.

Ah - right - of course. You have allowed for it in the determination of your expected spread - I should have realised that. Yep - it is just another outcome really so that it figures as a loss just the same as the opposite EC coming out would do. Cool. Be interesting to see how it works out over time. Thanks.

Nice work, LF. 

My play is based around similar ideas - divide and conquer. 

I think you have all the elements in place here for a winner!