Randomness in Bet Selection and/or Number Selection?

[Nickmsi]

I have read most of the threads on this forum and can conclude from Albertojonas, John Legend, Gizmotron, Ego, Esioto and many others that RANDOMNESS is the key to success.

We are trying to achieve randomness in the bet selection process by creating 4 and 5 wide matrices, by alternating Red/Black with SOSSOOS (Same, Opposite), by triggers of a color change, by betting that a Quad won't become a 5 pointer, etc.

However, there are Two Phases to this game of Roulette.

1.  Number Selection
2.  Bet Selection

So far we seem to  have only addressed randomness in the Bet Selection process.   

What would happen if we added randomness to the Number Selection?  Would this make the Bet Selection even more Random or make no difference at all?

For example, the wheel or RNG selects 3 Blacks in a row.   We know that if we play EVERY 3 Blacks in a row, we will eventually see 10 consecutive Blacks, 15 consecutive Blacks or 20 or more consecutive Blacks.

What if 3 Blacks in a row were randomly selected, therefore, NOT EVERY 3 Blacks in a row would be played.   Would this reduce the number of dreaded long streaks or not make any difference?

I have added randomness to the number selection process and done over 100 tests.   But this sample is too small to see if it makes any statistical difference.   I was testing Double Dozens not repeating which has about a 67% chance of winning on first bet.  My test results are close to 70%.   Before continuing, I thought I would check with some of you to see if adding randomness to the number selection really makes any difference.

Perhaps, the number selection is already random and adding randomness changes nothing.

What do you think?

[Playborne]

I quite agree, but Albertojonas, John Legend, Gizmotron, Ego, esoito and many others claim that adding randomness changes odds in our favor...

[LuckyLucy]

Working with/against random, is something I had never thought of before joining this forum. Some of the Matrix used on here are superb. Plus with us all working to the same goal, we can come up with better and more fine tuned systems.

[Bayes]

Perhaps, the number selection is already random and adding randomness changes nothing.

BINGO!

Playing 'random vs random' doesn't work because all patterns are equally likely. 

You can't add more randomness to a game which is already random, and even if you could, why should it make any difference to your win rate?  Suppose you have an idea that betting alternately R B R B ... is going to stop long losing streaks because it will 'break up' streaks of colours like RRRRRRRR or BBBBBBBB. True, and it would work if those were the only kinds of sequences which appeared, but it should be obvious that you just as frequently get choppy sequences in which betting alternate R/B  will give more losses than wins. The game is perfectly symmetrical with respect to patterns which can appear, and random vs random bet selections don't take this into account.

[esoito]

True. True.

But in the short-term (i.e let's say, the hit'n'run scenario) others' results seem to indicate 'short-term clusters' are often profitable.

(That's if the results claimed are, in fact, not the product of creative, wishful thinking, of course.)

I sense a circular argument starting to develop on this vexed issue, as others possibly weigh in to the debate!   

[Bayes]

Well, the whole concept of hit 'n' run is flawed to my way of thinking. Yes, it's a fact that there are short term clusters, but playing 'random vs random' doesn't take any account of when those clusters might occur,  so if you start betting your random selections randomly (with regard to entry point) then how can it give any advantage?

You can't separate your sessions from the accumulated number of spins you play over a lifetime, so that makes a nonsense of the whole notion of hit and run. It might make sense if there were a way of picking your spots which gave an advantage - you would wait for the opportunity, play it and then wait for the next, but this isn't what I understand as 'hit 'n' run' as described by most people. Usually, it seems to imply that you will be avoiding losing streaks by not playing too long at any one session, but what's the difference between taking a short walk across a minefield every day for a week, and taking one long walk once a week? it's the mileage which counts, not the time you take to cover it.

[GLC]

Well, the whole concept of hit 'n' run is flawed to my way of thinking. Yes, it's a fact that there are short term clusters, but playing 'random vs random' doesn't take any account of when those clusters might occur,  so if you start betting your random selections randomly (with regard to entry point) then how can it give any advantage?

You can't separate your sessions from the accumulated number of spins you play over a lifetime, so that makes a nonsense of the whole notion of hit and run. It might make sense if there were a way of picking your spots which gave an advantage - you would wait for the opportunity, play it and then wait for the next, but this isn't what I understand as 'hit 'n' run' as described by most people. Usually, it seems to imply that you will be avoiding losing streaks by not playing too long at any one session, but what's the difference between taking a short walk across a minefield every day for a week, and taking one long walk once a week? it's the mileage which counts, not the time you take to cover it.

Here's my thought on Hit N Run.  I agree with you Bayes that to just walk up to a table and start playing for a short time and then repeat the next day etc... gives us no real advantage over just playing continuously.

But, if we were to track, betting virtual until we had a drawdown on paper that started to recover and then we start playing for real, we might be able to catch a recovery period and since we didn't lose anything during the losing period the recovery period will be a winning period.  Sorry about that last sentence, but I hope you know what I mean.

I have examined a lot of the systems I play and there is a definite undulation to  the losses and wins periods.  I acknowlege that you can't know when a losing streak will end, but you can use markers like if you would have just lost say 10 units in the last 20-30 bets and you have 2 or 3 wins in a row, you might be entering a recovery period, so jump in.  You may have to put another qualifier in, say if you lose 3 of the next 5 bets, stop playing for real and continue virtual, etc...

By picking our spots to play, we might be able to win more than we lose for short periods.

What do you think?

GLC

[Bayes]

George,

What you've described is pretty much the way I play. 

I track several fairly 'standard' bet selections (FTL, DBL, one side, and ZZ) looking for swings in one direction or another, then jump in. I never count on an immediate swing back; nothing more than an average win rate or even somewhat less is good enough with the MM I use. Often I flat bet the minimum waiting for the turn, then up stakes, depending on how much I've lost waiting for things to move in my favour.

It isn't foolproof, but as you say,  there is a constant ebb and flow; all I'm doing to trying to catch it and go with it. It's not what I'd call spot-betting either, because I track all 3 ECs and there's something happening pretty much all the time.

Am I committing the gambler's fallacy? maybe, but there is a superficially similar statistical phenomenon called 'regression to the mean' which says that if a variable is extreme on its first measurement, then it will get closer to the average on the second. Really, all it says is that a particular distribution will 'behave' as it should do in the long term, so an EC can seem like a 2/1 shot (in terms of wins/losses) for a while, but sooner or later it will revert back.
For example, suppose you were to see a 115 spin sequence in which only 40 reds appeared. This is 3 standard deviations from the mean. Chances are that the next 115 spins will see more than 40 reds, if it didn't, that would mean a standard deviation of at least 4.2 over the sequence taken as a whole. Now I concede that this doesn't give you an advantage mathematically, but it does tend to make catastrophic losses less likely (although not guaranteed).

What I do isn't quite as simple as perhaps it sounds, but that's it in principle. As you yourself suggested, I track over multiple 'spin frames';  it gives additional safeguards.

[GLC]

Thumbs up Bayes.

I like that way of playing.  I play all my systems similarly now.
I've been having much better results tracking losses vs wins instead of a certain pattern.

I'm also having good results playing mini games under a global pattern.  The global pattern can be as simple as +1 after 2 losses, -1 after 2 wins.  Each mini game is either a loss or a gain and is treated like a decision of 1 bet.

Cheers

George

[ZeroBlue]

@GLC


One can try to identify tendencies and indications to try catch on a correction period. But it is hard to get strong Ecarts.
suppose you get in on a window of ecart with -3,0. What can be an indication that it won't go -6,0?
Statistically, once you get one -3,0 ecart does it tend to zero (correction) or to -6,0? what occurs more often? Mathematically i guess there is no positive expectation...


@Nickmsi

There are many ways still to try out. One way of randomize number selection is to group past outcomes in an order and play against or in favor of them in the future outcomes.
One already very known way is to capture groups starting with the same outcome
Example 4 dozens:
Dozen
1
2
2
3
3
2
1
1
2
3
3
3
2
2
1
2
1
3
1
1
....




dozen 1 = Capture 122 - 112 - 121
dozen 2 = Capture 223 - 211 - 233
dozen 3 = Capture 332 - 333 - 311


Now whenever each dozen comes out you can play it against repeating the pattern you captured
Or
play it will repeat will get you the same result.


The point is if you play in clusters of two bets and observe the LW registry you can try to interact with a soft progression and come ahead. MM is of utmost importance here.
---------------------------------------------------------------------------------------------------------------------
As Bayes states the Mathematical odds are the same.


I once saw a system that played 10x dozen 1 and if not ahead it followed more ten times last coming dozzen adding one unit to the flat bet. It was surprisingly hard to sink.


Another variation was to follow the most hit dozen from last ten spins with same results.




Good Luck 

[Bayes]

One can try to identify tendencies and indications to try catch on a correction period. But it is hard to get strong Ecarts.
suppose you get in on a window of ecart with -3,0. What can be an indication that it won't go -6,0?
Statistically, once you get one -3,0 ecart does it tend to zero (correction) or to -6,0? what occurs more often? Mathematically i guess there is no positive expectation...

My research over millions of spins shows that it's extremely rare for any sequence to go over 5.00 SD, that's a worst-case scenario which you many not ever see, and the longer the sequence the rarer it becomes, as I showed in the previous post.

[ZeroBlue]

My research over millions of spins shows that it's extremely rare for any sequence to go over 5.00 SD, that's a worst-case scenario which you many not ever see, and the longer the sequence the rarer it becomes, as I showed in the previous post.

That is kind of good news...     It requires patience to profit from those. 


How many events in a window should be considered short / medium / long sequence?

[Nickmsi]

 Thanks all for your input and ideas.  You have saved me a lot of time and energy trying to make random number selection more random.

I was tracking in Excel a Casino number followed by a RNG number from Excel (RandomBetween (1,36) followed by another Casino number, then another Excel random number, etc. .  I was testing bet selections based on repeaters or sleepers from these alternating random numbers.

Random is Random.  Infinity is Infinity.  They are what they are. 

Thanks again, it was a fun experiment and learning process. By the way, my test results were as expected pretty close to the statistical norm.

Nick

[hanshuckebein]

@bayes

I totally agree with you that two combined truly random processes don't create something like "superrandomness".
still I've read that roulette with its 36 numers plus zero does not fullfill the requirements to be truly random.

to be truly random and to have things equally distributed on the layout and the wheel as well, it must consist of 48 numbers plus zero.

I really would like to hear your opinion on this calculation. 

cheers

hans

[Bayes]

Hi hans,

Where did you see that calculation and can you give some details? On the face of it, I can't see how any such claim can be justified - all numbers are equally likely to hit, and that fulfils the requirements of random.