### Author Topic: Where is the proof that variance can be reduced?  (Read 56613 times)

0 Members and 6 Guests are viewing this topic.

#### Mike

##### Where is the proof that variance can be reduced?
« on: October 19, 2015, 12:26:18 PM »
A lot of system players argue that it's not the house edge that makes you lose, but the variance -- the wild swings and bad runs.

I concede that IF (and it's a big if) you could reduce the variance, then some kind of progression might work. If it could be reduced enough, then it certainly WOULD work.

Trouble is, the only way to reduce the variance is by increasing the accuracy of your predictions, and since spins are independent, past spins don't indicate future spins. IF you could reliably forecast a series of even money bets such that the longest losing run was 3 or 4, or even 5 or 6, then you could clean up with a simple double-up progression. In my view this is a fantasy; there is no way of reducing the variance to anywhere near such a level.

I'm aware, of course, that one of the ways of reducing volatility in stocks and shares is to spread your risk over a number of markets, but I don't see how this can be applied to roulette. The concept of hedging can't be applied in a random game.
« Last Edit: October 19, 2015, 02:33:48 PM by Mike »

#### dobbelsteen

##### Re: Where is the proof that variance can be reduced?
« Reply #1 on: October 19, 2015, 12:48:01 PM »
Variance can help the stratistic player not the system player. The House edg is only valid for Long run events.
All roulette players , included APs, play short run sessions. Short run events have the the feature they can lose or win. Hit and run is the sollution.

#### Mike

##### Re: Where is the proof that variance can be reduced?
« Reply #2 on: October 19, 2015, 02:45:36 PM »
Quote
Hit and run is the sollution

dobblesteen,

I disagree.

For hit and run to be effective you need to (a) know that there ARE favorable opportunities, and (b) Know when they occur, so you know when to hit and when to run.

Think about card-counting in Blackjack. The proficient card-counter knows the theory and how to count the cards, so he will "hit" when the count favors him and "run" when it doesn't. But in the game of roulette there can be no such "triggers" (notice I say "game". This discounts AP).

The only indicators that the system player can "rely" on are based on the faulty logic that past spins indicate future spins, so every so-called "opportunity" is really just a guess. There are no objective indicators for when to hit and when to run.

Unless you're saying that playing short sessions is in itself sufficient to give you an advantage? But of course that would be absurd. It may give you the illusion that you are losing less simply because it takes you longer to play the same number of spins.
« Last Edit: October 19, 2015, 03:03:20 PM by Mike »

#### kav

• Hero Member
• Posts: 2367
• Thanked: 1327 times
• Gender:
##### Re: Where is the proof that variance can be reduced?
« Reply #3 on: October 19, 2015, 02:56:48 PM »
Mike,

Yes, I do believe it is the variance that kils the player not the house edge. So you thread is to the point. I give you that!

#### dobbelsteen

##### Re: Where is the proof that variance can be reduced?
« Reply #4 on: October 19, 2015, 04:58:58 PM »
Reduced is not the same as killing.

#### Bebediktus

##### Re: Where is the proof that variance can be reduced?
« Reply #5 on: October 20, 2015, 06:19:46 PM »
Quote
All roulette players , included APs, play short run sessions. Short run events have the the feature they can lose or win. Hit and run is the sollution.
No dobbelsteen, if i find good wheel, where i can have edge, i play as long as i can, because longer i will play - more i will win.
Of course can be that after long play i become tired and begint to do mistakes , but that is other story.

#### Real

• Fighting the war on absurdity one foolish idea at a time.
• Hero Member
• Posts: 1684
• Thanked: 283 times
• Gender:
##### Re: Where is the proof that variance can be reduced?
« Reply #6 on: October 20, 2015, 06:42:27 PM »
Variance can help the stratistic player not the system player. The House edg is only valid for Long run events.
All roulette players , included APs, play short run sessions. Short run events have the the feature they can lose or win. Hit and run is the sollution.

That's not true.  We try to play as long as possible so that our edge will dominate the variance and enable us to win as much as possible.  We aren't trying to win just a few units here or there....we're playing to win several hundred units each time we play.
« Last Edit: October 20, 2015, 06:44:08 PM by Real »

#### sqzbox

##### Re: Where is the proof that variance can be reduced?
« Reply #7 on: October 21, 2015, 12:15:24 AM »
Reduced?  Mmm - not sure it can be. I'd love to see some real evidence of this also. But can it be "managed"? To your advantage of course. I guess the AP guys will say that their techniques do it - but does it really? Given that you are playing the physics of the wheel I'm not sure that random variance, in a statistical sense of the random game, is an appropriate measure for you.  You would have your own variance measured on your game which would be different I suspect to that of the mathematical model of a random game.

Actually, as I think about this (great question Mike by the way!), I begin to wonder what "reduce the variance" actually means. Isn't variance just an outcome from a particular binomial curve? Just like a bunch of other mathematical measures or results that can be calculated from data. And that binomial is determined from actual data - yes? So is the question actually more about the management or control of a variance that, itself, cannot be changed (since it is just a calculation from the data at the end of the day).

The more I think about it the more I think that the question is perhaps posed as a trap, or a joke. Because - variance is just a result of a particular curve and that curve results from data, and that data comes from a play - a play strategy I mean. So if you change your strategy then the data collected will be different from other strategies and the variance for that particular set of data will be different to the variance from another set of data where the selection strategy is different again.

Am I splitting hairs? Perhaps. But I wonder if the question is more - can variance be managed in such a way as to allow a progression to succeed. By "succeed" I mean "never fail". A progression only has to fail once for it to be useless long term. Or perhaps the question is posed as it is deliberately to allow the system players to bash their collective heads against an imaginary brick wall for a while.  Nice one Mike!
« Last Edit: October 21, 2015, 12:17:48 AM by sqzbox »

#### sqzbox

##### Re: Where is the proof that variance can be reduced?
« Reply #8 on: October 21, 2015, 12:24:33 AM »
As regards the suggestion of hit and run - I've never understood why people continue to put this approach forward as a viable one - one that somehow magically defeats the math. Maybe I don't understand what they are getting at - perfectly happy to have my understanding increased so have at it! But, don't you guys understand that there is absolutely no difference, mathematically or philosophically, between one long session of 10000 spins or a combination, end to end, of 100 games of 100 spins? None. Nada. Zip. So how can hit and run be different?

#### Trilobite

##### Re: Where is the proof that variance can be reduced?
« Reply #9 on: October 21, 2015, 02:53:48 AM »
As regards the suggestion of hit and run - I've never understood why people continue to put this approach forward as a viable one - one that somehow magically defeats the math. Maybe I don't understand what they are getting at - perfectly happy to have my understanding increased so have at it! But, don't you guys understand that there is absolutely no difference, mathematically or philosophically, between one long session of 10000 spins or a combination, end to end, of 100 games of 100 spins? None. Nada. Zip. So how can hit and run be different?

I've disproved hit and run for myself and have tried many times to explain why it is ineffective. Its like flogging a dead horse.

Once you know how to stabilise the random variance of a particular data set, you can effectively employ a progression to suit the common random outcomes within that data set.

Stabilising the random variance of a particular data set means increasing the number of bets placed on the more common random outcomes and decreasing the number of bets on the uncommon random outcomes

There will always remain some uncommon random outcomes, so the progression needs to be fine tuned to the data set so that occasional losses or progression busts do not cause the system to return to the negative expectation of the original random variance of the data set.

It is extremely hard to prove whether or not a system can achieve this control over the original variance.
« Last Edit: October 21, 2015, 02:56:17 AM by Trilobite »

#### Mike

##### Re: Where is the proof that variance can be reduced?
« Reply #10 on: October 21, 2015, 07:56:18 AM »
sqzbox,

The more I think about it the more I think that the question is perhaps posed as a trap, or a joke. Because - variance is just a result of a particular curve and that curve results from data, and that data comes from a play - a play strategy I mean. So if you change your strategy then the data collected will be different from other strategies and the variance for that particular set of data will be different to the variance from another set of data where the selection strategy is different again.

I wasn't posing the question as a trap or joke; theoretically there ARE ways to reduce variance in some fields of speculation, I just doubt whether it can be done in the random game of roulette.

Changing your strategy will change the way the wins and losses are distributed, but it shouldn't change the variance. If you bet on the even chances and use the classic "before last" bet selection then the wins and losses will be distributed differently than if you use the "bet opposite to last" way of selection, but the variance for both will be the same in the long term -- the GAPS between wins will average out to the same length in both cases. If you were able to reduce the gaps such that the average gap was smaller then you would have reduced the variance.

But the only (?) way to reduce the average gap length is by increasing the accuracy of your predictions, and how do you do that in a random game when all you have to go on are past spins (which don't indicate future spins).

Trilobite,

Quote
Stabilising the random variance of a particular data set means increasing the number of bets placed on the more common random outcomes and decreasing the number of bets on the uncommon random outcomes

What do you mean by "more common"? This sounds like trending -- so bet on the numbers which are "hitting the most", yes?

Well in that case you're back to the problem of independent trials. But I agree that's better than betting against the "flow" or on cold numbers if only because if you do that you might be betting against some bias. Better to bet with the bias than against it. ;-).

#### sqzbox

##### Re: Where is the proof that variance can be reduced?
« Reply #11 on: October 21, 2015, 08:34:56 AM »
Totally agree Mike. But what I meant by changing play strategy was on a different bet type. Like EC's might be one but betting straight-up might be another. Or perhaps 2 columns at a time. I agree that ALL strategies on the EC's will have the same variance. But in any case, that is not the point of the discussion.

I suspect that Drazen would argue that he has effectively overcome the variance in his approach - of waiting for the extreme event and then running a short hard-hitting progression. Personally, I'm not convinced, but I guess it is hard to refute his results - if they are indeed as good as he says. But what would be your response to a strategy that hits the statistics? Let's say, for example, that the longest run seen to date of an EC is 23. So you wait for 20 and then play a 5 step marty (to be safe this allows a couple more than the expected maximum). Is this a winner? Let's not worry about the reality of waiting for such an event (probably years) because this is a theoretical discussion right? So if we were to say that this is a successful strategy then - exceptio firmat regulam
and so we have an approach.

Other than that I would have to agree that the only way is to increase the accuracy of your selections. In other words, somehow defeat probability. Could that be done by utilising the stats of the probability engine perhaps? Waiting for a combination of extreme events and then playing the normality afterwards?

#### Trilobite

##### Re: Where is the proof that variance can be reduced?
« Reply #12 on: October 21, 2015, 09:54:21 AM »

Trilobite,

Quote
Stabilising the random variance of a particular data set means increasing the number of bets placed on the more common random outcomes and decreasing the number of bets on the uncommon random outcomes

What do you mean by "more common"? This sounds like trending -- so bet on the numbers which are "hitting the most", yes?

Well in that case you're back to the problem of independent trials. But I agree that's better than betting against the "flow" or on cold numbers if only because if you do that you might be betting against some bias. Better to bet with the bias than against it. ;-).

By more common I mean placing bets that are within the expectation of the bet type.
If the bets donâ€™t resolve within the expected value then you look for another bet rather than pushing on with the original bet.

This is a baby step toward gaining stability over variance.
« Last Edit: October 21, 2015, 09:55:56 AM by Trilobite »

#### dobbelsteen

##### Re: Where is the proof that variance can be reduced?
« Reply #13 on: October 21, 2015, 10:06:10 AM »
Hit and run has two faces. You can stop after a profitable session or leave the casino after a profitable visit.
I visit my house casino about 5 times a week. Normally I play 2 to 3 hours.
I stop playing when I have a profit of 100 euros or I stop playing after a small profit in the last halve hour to avoid a loss at the end of my visit. I am very satisfied with a small profit. On the first place I visit the casino for pleasure.
Every visit is a short run event. Everybody has a chance of 50% on a win during a short run session.

#### Bebediktus

##### Re: Where is the proof that variance can be reduced?
« Reply #14 on: October 21, 2015, 10:40:39 AM »
I stop playing when I have a profit of 100 euros or I stop playing after a small profit in the last halve hour to avoid a loss at the end of my visit. I am very satisfied with a small profit. On the first place I visit the casino for pleasure.
In last sentence we are very diferent , really i not very like casinos , i can find what to do other with big pleasure, but problem is money - casinos is mine job and i not know how to get money in other way - only how to win them from casino. So i cant go to casino for 100 eur, that for me simply inefective . If i could play every day 8 hours then for me maybe will be enough 100 eur /hour , but i cant.

So when i see that in near future i will win , i simply cant stop and go out.

I understand - your win is more like gift for you - you not know when you will win when not. I that know. In your place maybe mine behaviour will be the same - won 100 and run out and never return