728x90 welcome bonus

Author Topic: In Defense of Patterns  (Read 2793 times)

0 Members and 1 Guest are viewing this topic.

Bayes

  • Veteran Member
  • ****
  • Posts: 688
  • Thanked: 550 times
  • roulettician.com
In Defense of Patterns
« on: October 06, 2016, 10:53:50 AM »
In gambling circles players are often accused of finding meaning in patterns where none exist. This is called Apophenia:

Apophenia is the human tendency to perceive meaningful patterns within random data.

Examples include gambler's fallacy and overfitting (according to the wikipedia article on Apophenia).

But "meaning" has to have a reference - meaning with respect to what? In roulette, it could be said that the pattern RRBRRB is meaningless with respect to the physics of the wheel, and that it is meaningless because it's just one pattern among 64 patterns which are equally likely. But no pattern is inherently meaningless, because we have to find our own meaning in patterns. The very fact that there are 64 different patterns in 6 spins and that "random" itself has a pattern suggests that any particular pattern cannot be meaningless.

Even though such patterns may not be very good predictors of other patterns not yet observed it doesn't mean that they can't be profitably used, or that they are "meaningless".

And in fact, the gambler's fallacy isn't really about "finding meaning in patterns where none exists", it's about not understanding that a "fair" wheel means outcomes are independent as well as unbiased. It is not a fallacy to make decisions based on data collected, despite what some AP dogmatists keep telling us.

Machine learning is largely about finding meaning in patterns.

Quote
Within the field of data analytics, machine learning is a method used to devise complex models and algorithms that lend themselves to prediction - in commercial use, this is known as predictive analytics. These analytical models allow researchers, data scientists, engineers, and analysts to "produce reliable, repeatable decisions and results" and uncover "hidden insights" through learning from historical relationships and trends in the data


 
The following users thanked this post: kav, palestis, Sputnik, scepticus, Reyth, Sheridan44

kav

  • www.Roulette30.com
  • Administrator
  • Hero Member
  • *****
  • Posts: 1830
  • Thanked: 801 times
  • Gender: Male
Re: In Defense of Patterns
« Reply #1 on: October 06, 2016, 11:41:14 AM »
Thanks for this post Bayes.
These subjects are of great interest to me.
Roulette is so much more than throwing a few chips on the table and hoping to win. There is so much more to this game and how we perceive and interact with it.
 
The following users thanked this post: scepticus, Reyth, MrPerfect.

scepticus

  • Hero Member
  • ******
  • Posts: 2040
  • Thanked: 421 times
Re: In Defense of Patterns
« Reply #2 on: October 06, 2016, 01:38:24 PM »
Thanks for this post Bayes.
These subjects are of great interest to me.
Roulette is so much more than throwing a few chips on the table and hoping to win. There is so much more to this game and how we perceive and interact with it.

I think all gambling is about perception -  how we look at things.
Like most  members here I regard roulette as a challenge  .A challenge to profit from guessing the result/s of  future spins which are unknowable beforehand. A challenge to prove naysayers wrong.
Too true, kav,there is much more to roulette than throwing chips on the table and hoping to win !
 
The following users thanked this post: BlueAngel, Reyth, MrPerfect.

MrPerfect.

  • Great Contributor
  • *****
  • Posts: 978
  • Thanked: 542 times
Re: In Defense of Patterns
« Reply #3 on: October 06, 2016, 03:31:14 PM »
 So here about data. Then we speak about data and predictive analytics,  we speak about forcast of model behaviour.  First we need to collect data that helps us to model. My proposition is to model physical conditions and their affect on outcome. To model conditions , first we have to define them.
   Object of predition in this case will be likehood of outcome (final number), depending on these conditions.
   From practice l found these conditions being able to correlate with final number:
   Rotor speed
   Ball in use
   Direction of the spin
   Diamond hit... ets.
 All these conditions are phisical variables wich can be taken during the spin. Even presence of these conditions in last "x" spins can be correlated to outcome. In this case we look how likely nessesary conditions are to be served In the next spin/ spins.
  Assuming we are able to use multivariable correlation, we may make crosstest between sessions, days, randomly selected groops of events within our data sample to determine boundaries wich still produce positive result for us. It will give us data on wich we can run simulation of different situations in play to profit optimally. 
     Now , if we skip all physics data taken and focus on results only, or randomly selected groups of results ( ec, dousens,  6lines .. est) , how to correlate such results with next results?
 Obviously we can observe patterns even there, but wich level of control do we have in this situation?
   In other words , where to put money and wich return to expect in the long run?

 
The following users thanked this post: Reyth

Reyth

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 3530
  • Thanked: 1091 times
Re: In Defense of Patterns
« Reply #4 on: October 06, 2016, 11:40:05 PM »
Wow Mr. Perfect!  Welcome to the forum!

Like you Scep, I want to know the results of the upcoming spins PRIOR to their arrival!

Unlike Mr. Perfect (and everyone else reading this, blush) I base this on the statistics concerning the random flow (or like Dobble calls it, the "random rows").  I do this based on the results generated over many millions (or even billions and greater) of spins; at a certain point, overall results do not substantially change (except for minor and normally occurring variance) and I am now calling this:

The Practical Long RunTM

Our goal as gamblers is to find the ways that maximize our ability to access & understand the worst sequences that we can expect that are generated by these spins because if we are able to defeat and/or avoid these sequences, we have practically defeated roulette itself.

From my experience, not much that we can do makes a difference in the long term unless we first take a long term view and that requires deliberate intention.

A problem exists when people tell us "its impossible", "there is no way to change anything" and like statements.

Here is an example from business life:

Over 96% of businesses fail within 10 years.  Tax revenue is an inescapable constant(1), whereas business revenue is only relatively certain(2) and so eventually the tax revenue will eclipse the business revenue and your business will fail.  Because of the principles (1) & (2), there is no need to engage in business.

Obviously, the example above is not exactly the same as the arguments against winning in roulette but it is similar and exemplifies how much of an "easy way out" a defeatest attitude is. 

Company loves misery I guess?   
« Last Edit: October 07, 2016, 12:21:09 AM by Reyth »
 
The following users thanked this post: MrPerfect.

MrPerfect.

  • Great Contributor
  • *****
  • Posts: 978
  • Thanked: 542 times
Re: In Defense of Patterns
« Reply #5 on: October 07, 2016, 12:36:04 AM »
Wow Mr. Perfect!  Welcome to the forum!

Like you Scep, I want to know the results of the upcoming spins PRIOR to their arrival!

Unlike Mr. Perfect (and like everyone else reading this, blush) I base this on the statistics concerning the random flow (or like Dobble calls it, the "random rows").  I do this based on the results generated over many millions (or even billions and greater) of spins; at a certain point, overall results do not substantially change (except for minor and normally occurring variance) and I am now calling this:

The Practical Long RunTM
Our goal as gamblers is to find the ways that maximize our ability to access & understand the worst sequences that we can expect that are generated by these spins because if we are able to defeat and/or avoid these sequences, we have practically defeated roulette itself.

From my experience, not much that we can do makes a difference in the long term unless we first take a long term view and that requires deliberate intention...
I like your trade mark. :)
  OK,  let's assume that no one jumps into long run from one moment to another. I mean no one will play 5k spins without a break for a siggarete.
   Anyhow, wheels do have their " personality" , some numbers do fall more then others , be it group of continuous numbers or separate numbers selected according to some criteria.
 I think that nessesity of wise selection of numbers is obvios- they fall more often, so more chances to win.
   However, l do not see many people to make " systems" , " rools", " triggers" for situation of wisely selected numbers.
   I would be very curios to know if you ever considered situation where you actually do have an advantage and may expect your numbers to hit more often then simplistic probability dictates.
   Up till now l run some simulations on real datasets to determine empirically where boundaries of abuse in betting are. Obviously these simulations do represent " most common" situations only and do have in fact much room for improvement. ...
   Could you populate please how did you do your simulations and if option of " known advantage" could be implemented ?
  My areas of interest in such simulations would be :
 Simulate jump into the game in different moments, likehood of worst case scenario, ability to set up triggers to recognise when bet more or when bet less.
 
 

Reyth

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 3530
  • Thanked: 1091 times
Re: In Defense of Patterns
« Reply #6 on: October 07, 2016, 01:17:25 AM »
Honestly, the best that I can say I have discovered is Harry-Pales' Betting Range TechniqueTM

I have attempted to lock down a concrete advantage by obtaining a list of the most favorable numbers but only to have it nullified by the "statistical features of the random flow", otherwise known as, "the reverse engineering curse".

So I decided to give up on trying "lock anything down" in the direct manner but found I was able to address the advantage I desired indirectly, by using a successive, FAVORABLE RANGE of bets.

The method consists of:

1) Determine your desired bet selection(s)
2) Output the long term loss distributuion, showing the percentage of hits versus losses.
3) Choose a favorable betting range that consists of: a) trigger = number of consecutive misses, b) progression length = number of desired bets based on the favorable percentage chances of gaining a hit
4) Run the trigger/range parameters in a simulation which tracks the max consecutive successive losses per "level"

If your trigger was 10 spins (consecutive misses) and your betting range was 5 spins you would enter the loss distribution at spin 11,12,13,14 & 15 which have their own favorable statistics.  There is a practically expected limit to the number of times successively that this range will lose before no more consecutive losses appear.

I have been able to take an EC with a maxmimum expected consecutive loss of 25 and reduce it down to only 11 by using this technique (a trigger of 7 and a betting range of 9); level 1 is bets 1-9 and level 2 is bets 10 & 11.

Changing the trigger/range figures affects the total number of expected consecutive losses!!
« Last Edit: October 07, 2016, 01:19:31 AM by Reyth »
 

MrPerfect.

  • Great Contributor
  • *****
  • Posts: 978
  • Thanked: 542 times
Re: In Defense of Patterns
« Reply #7 on: October 07, 2016, 02:29:56 AM »
 Reyth, what kind of simulator do you use? Is it montecarlo method? Would you be able to program it with desired " how many times each EC hits in your dataset"?
 

Reyth

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 3530
  • Thanked: 1091 times
Re: In Defense of Patterns
« Reply #8 on: October 07, 2016, 04:34:40 AM »
Reyth, what kind of simulator do you use? Is it montecarlo method? Would you be able to program it with desired " how many times each EC hits in your dataset"?

I program in BASIC which gives me 100% freedom on what I am able to create.

What is he Monte Carlo Method?  Do you mean the Martingale?

I believe your last question means "number of appearances of an EC in the last X spins" in which case you wish to "loop the betting line into a circular analysis" for ongoing repeat betting?  If so, I think it should be done it two stages, where first our theory about the aggregate appearances is proven and then the monitor is created to suggest the bets based on that method.

I have the statistics for a virtual EC (18 inside numbers) and their minimum appearances for a group of 1-616 spins.

Currently I have only actively bet upon streaks of EC's from the outside & a virtual dozen (2 DS) from the inside and have not bet on aggregates.
 

MrPerfect.

  • Great Contributor
  • *****
  • Posts: 978
  • Thanked: 542 times
Re: In Defense of Patterns
« Reply #9 on: October 07, 2016, 08:15:19 AM »
 Montecarlo simulator method is to use RNG to create data streams over and over, with given parameters.  Normally used to assess risk, kind of " what if" analysis simulator.
   It permits to see how likely are different scenarios and is very useful to create optimal betting model.
   You could read about it here   http://www.palisade.com/risk/monte_carlo_simulation.asp
   
  Area of common interest would be to run such simulations for given amount of numbers ( not nessesary 18, or as you call it " virtual EC".).
   
« Last Edit: October 07, 2016, 08:21:35 AM by MrPerfect. »
 
The following users thanked this post: Reyth

Bayes

  • Veteran Member
  • ****
  • Posts: 688
  • Thanked: 550 times
  • roulettician.com
Re: In Defense of Patterns
« Reply #10 on: October 07, 2016, 08:48:05 AM »
  Obviously we can observe patterns even there, but wich level of control do we have in this situation?
   In other words , where to put money and wich return to expect in the long run?

Good question. The truth is, I don't know. But what I do know is that my results are better when making use of past results than when I'm betting randomly, or just ignoring them. By "better" I don't necessarily mean making a flat bet profit, but the variance is definitely reduced. Currently I'm just using my "intuition" to select bets (not a mechanical system), but this is unsatisfactory to me. I would like to be able to devise a 100% mechanical system which performs at least as well as my intuition. The techniques of machine learning might help, or maybe not, but it'll be fun trying.  :)

In the context of machine learning, perhaps it's the case that you would be using supervised learning (because you have a list of variables which correlate with the outcomes), whereas for purely statistical research (no physical variables) unsupervised learning might be more appropriate. Of course, data in itself doesn't "cause" other data in the way that physical variables cause (or are correlated with) definite effects such as bias, but there may be some regular patterns or clusters which are associated with other "events". 
 
The following users thanked this post: Reyth

MrPerfect.

  • Great Contributor
  • *****
  • Posts: 978
  • Thanked: 542 times
Re: In Defense of Patterns
« Reply #11 on: October 07, 2016, 09:16:52 AM »
@Bayes, you have Real's book, don't you?
  If you wish, add me on Skype, l will show you how to resolve " plane bet " problems.
   Real adventure starts after resolving it, determine best betting strategy for example, optimal betting, triggers to call you into game or to stop betting.. ets.
  You are probably more " math wise" then me, so your expirience could be beneficial for both of us, and probably for community in general. Let's try to cooperate?
 
The following users thanked this post: Reyth

scepticus

  • Hero Member
  • ******
  • Posts: 2040
  • Thanked: 421 times
Re: In Defense of Patterns
« Reply #12 on: October 07, 2016, 09:31:41 AM »
There are always patterns, aren’t there  ? Any consecutive three spins of EC , for example, must be one of only 8 patterns ( excluding zero ).
My focus is on the short term patterns of 3 (EC) or 4 (Dozens/Columns) and  a process of elimination.
I see no need for long trials which can be trashed by variance.To my mind they are only good for giving “ confidence” to your method - and good luck with that. 
Progressions ? .Flat Bets are fine by me so why should I engage in progressions which require a larger bankroll ? I agree with HarryJ and Pally -reduce the bankroll needed  because recovery after a loss is more speedy.
The Bottom Line for me is -” Does your Method-whatever- show an overall  profit “? All the rest is “ noise”- entertaining and thought  provoking  -  but still “noise“.
A minority view though. !
 
The following users thanked this post: Reyth

MrPerfect.

  • Great Contributor
  • *****
  • Posts: 978
  • Thanked: 542 times
Re: In Defense of Patterns
« Reply #13 on: October 07, 2016, 10:29:24 AM »
There are always patterns, aren’t there  ? Any consecutive three spins of EC , for example, must be one of only 8 patterns ( excluding zero ).
My focus is on the short term patterns of 3 (EC) or 4 (Dozens/Columns) and  a process of elimination.
I see no need for long trials which can be trashed by variance.To my mind they are only good for giving “ confidence” to your method - and good luck with that. 
Progressions ? .Flat Bets are fine by me so why should I engage in progressions which require a larger bankroll ? I agree with HarryJ and Pally -reduce the bankroll needed  because recovery after a loss is more speedy.
The Bottom Line for me is -” Does your Method-whatever- show an overall  profit “? All the rest is “ noise”- entertaining and thought  provoking  -  but still “noise“.
A minority view though. !
"noise" can be useful in the situation where we already are able to achieve positive result with plane bet.
 Optimum betting is plane for maximum of table. Providing that we already win with plane bet.
 But plane betting option is good only when our betting unit is high, in case of limited bankroll we need to consider other betting options avaliable , wich permit to elevate our bankroll antil we are able to profit properly with plane bet to the maximum. Ones bankroll requirements achieved,  we always can ( and should) to switch to plane bet in order to profit optimally.
 

dobbelsteen

  • Hero Member
  • ******
  • Posts: 1363
  • Thanked: 344 times
Re: In Defense of Patterns
« Reply #14 on: October 07, 2016, 10:47:11 AM »
MrPerfect what do you mean with the more fallen numbers. Is thet related to a wheel or a RNG.

What are your triggers playing  your method. I thought that AP players do not use triggers. An AP player decides his bets on observation of the wheel rotation and the velocity of the ball.

Reyth all 18 numbers bets has the same features, including the ECs.

Different long run flows of the same chance have no different features.