Neural networks

[insidebet]

To all the computer wise-guys...

I have been on VIP and now this site on and off for the past three years.  I have read for the most part without interfering too much.  At least 99% of the guys here are bright and well intentioned. (The other1% you know who you are).
I have spent way too many hundreds of hours testing systems from this site and my own.  I have come to a few conclusions and I also have a project that could be put forward by someone more capable than me.
Conclusions:
1- Waiting for an event to happen before betting is a waste of time.
Spins are truly independant. Therefore "hot" or "cold" is irrelevant.
2- Progressions will only make a losing system lose more.  They only create an illusion. The session from hell always comes to wipe any gains out the window, and more.
3- Money management is also a fallacy.  It is just about timing and nobody can predict good and bad streaks. 
4-A system needs to be tested at least 100k spins before you have any idea.  I have not RXed one that hasn't tanked by then.  These systems are very much like everything I read on this forum.  So, if you don't agree with me, it is your right.  But please don't come screaming at me that I have not tried this or that.  I HAVE!

But, I keep trying...  Trying is fun anyway.  I am most interested in new ideas.  GUT is a great example.  Maybe there is something in numerology, maybe not...

I did mentioned something really different two months ago.  Nobody picked up on it.  Maybe Alarian did.  I have read a couple of post he wrote that talked about something very similar.

The idea I have has to do with Neural Networks.

Maybe, just maybe if there is some sort of order in all the chaos, a powerful computer could see it.

Imagine the scene: you sit down with with your computer friend, have a bear with it, and say: "Mate, I have a favor to ask you. Here is ten million spins. Here are all the rules of the game of roulette.  I want you to play until you come up with a method that can make me win after at least 1 million bets. So there! Have fun and I will see you in a couple of months..."

Whether this would produce something good or not, I don't know.
Does anybody think it is worth the try?

Insidebet

[VLSroulette]

Hello dear InsideBet. I too find interesting the neural network concept.

By neural network do you mean a matrix-like arrangement of numerical events and perform picks based on a continually evolving historial data?

Thank you.

[pins]

if the spins are indepentant. what you feed into the computor is of no value. DO YOU THINK THESE SPINS ARE GOING TO COME AGAIN IN THE SAME ORDER.

[metalrat]

Hi Insidebet,

This last year I have done a lot of work with neural nets to try to find a playable system. Many hours of work I can tell you. At the moment I am working closely with esoito to find a working system based on numeronomy.
I have fed my neural nets only prepared live data from Dublin Bet. Approximately 13000 spins from different dealers. to be short the results I had were sometimes brilliant - 8 direct hits on single numbers in 100 spins - and sometimes rubbish, 300 spins without a single hit.
I have a theory that if you could collect 7000 spins from one dealer and subsequently use the trained net only when he or she was spinning, you may have great success.  Perhaps when a bot is used to harvest the DB spins and the dealers spinning, this may become reality. I am convinced that there are many elements a neural network can pick up on when one person is dealing - physical characteristics not noticeable to us.

Anyway, this describes my work in short.

cheers,

metalrat

[insidebet]

VLS,
A neural network tries to emulate the human brain in the sense that it will try to win through trials and errors. 
It is a bit like a new born baby at first: it doesn't know anything at all. Just the rules of the game itself.  It will make billions of calculations, permutations trying to detect "patterns".  That is exact thing most of us are trying to do here. 
If there is a Grail involving patterns, a computer should be able to find it.

But be careful.  It has to involve a very large number of possible bet to be meaningful.
I will give a very simple example.  Suppose your computer friend comes up with: if you see numbers 5 34 17 12 and 6 in a row then you bet number 9 and 35 for three spins.  Well!  That is very nice but this situation might have happened ten times in 10 million spins.  So this would be totally irrelevant.


Pins,

Like I said in my opening post, it appears that all spins are independant. It appears that way to the human eye and brain.  But there is a whole bunch of scientists out there who think that what appears to be totally chaotic is not really.
That is what a Neural tries to do: find some order through the chaos.

Metalrat,

I have two real rats as pets...Very intelligent, I must say.

Nice to see that some have already worked the field.
Would you trust the data from Wiesbaden?  I downloaded 2 500 000 spins from there for RX.
You could take dealer changes into consideration, but I don't think that would be necessary for neural research.  As I said, a large sample of data is much better

Regards

Insider

[Number Six]

I have a theory that if you could collect 7000 spins from one dealer and subsequently use the trained net only when he or she was spinning, you may have great success.  Perhaps when a bot is used to harvest the DB spins and the dealers spinning, this may become reality. I am convinced that there are many elements a neural network can pick up on when one person is dealing - physical characteristics not noticeable to us.

This is interesting, Metalrat. I have thought about this before, and I'm really glad to see that someone is testing. If neural networks can find no discernable patterns in such small samples of independent trails, then perhaps they can be found after considering the present human variable. To work this would mean studying the same dealer and analysing a massive amout of his/her spins. Of course, it would all be undone if the dealer quit the casino or got sacked or whatever! While I'm not convinced of the dealer signature idea and that it could be exploited from spin to spin, it's perfectly reasonable to think that a dealer will produce a repetitive rhythm throughout a whole session...now, this could be exploited if it can be picked up early enough. As far as roulette software robots go, this is where the future is.

[gizmotron]

VLS,
A neural network tries to emulate the human brain in the sense that it will try to win through trials and errors. 
It is a bit like a new born baby at first: it doesn't know anything at all. Just the rules of the game itself.  It will make billions of calculations, permutations trying to detect "patterns".  That is exact thing most of us are trying to do here. 
If there is a Grail involving patterns, a computer should be able to find it.

Anyone that tries can outperform the computer. The human can see similarities that are more like characteristics of form rather than the actual recurrence of an exact duplicate situation. Knowledge of characteristics, learned from practice, are more valuable than simple pattern recognition.

But be careful.  It has to involve a very large number of possible bet to be meaningful.

You can discern some characteristics and changes in just a few spins, if you have learned how to read randomness that is. You can know something about the situation at spin fifteen out of thirty of the full duration of a known trend type. I noticed a situation that lasted for 40 spins last night at the casino. I hammered it to death vigorously.

[bombus]

You can discern some characteristics and changes in just a few spins, if you have learned how to read randomness that is. You can know something about the situation at spin fifteen out of thirty of the full duration of a known trend type. I noticed a situation that lasted for 40 spins last night at the casino. I hammered it to death vigorously.

*.......sigh*

[gizmotron]

*.......sigh*

So why can't you read randomness? It's so easy a caveman can do it.

[bombus]

So why can't you read randomness? It's so easy a caveman can do it.

"Ug ug argh ug."

That's caveman for, "Yeah, well you can't read my big juicy bankroll balance, so we're even."

[gizmotron]

"Ug ug argh ug."

That's caveman for, "Yeah, well you can't read my big juicy bankroll balance, so we're even."

I'm convinced that you are not a successful roulette gambler. You make all the noises of a person that is in a constant search mode. You have never shown me that you have a core value here on this forum. So I don't need to respect your posts anymore. They are all just buzzing noises that serve nothing. It's too bad too. I thought you were going to be one of the smart ones. You've earned the right to be ignored.

Please stop being querulous.

[bombus]

I'm convinced that you are not a successful roulette gambler. You make all the noises of a person that is in a constant search mode. You have never shown me that you have a core value here on this forum. So I don't need to respect your posts anymore. They are all just buzzing noises that serve nothing. It's too bad too. I thought you were going to be one of the smart ones. You've earned the right to be ignored.

Please stop being querulous.

Well I can't argue with most of that.
A man's got to be true to his own convictions so I will leave you to yours.

As for the querulous snip, all I can say is...

Il bue che dice cornuto all'asino!

[bombus]

-