Quote: Originally posted by Guru on September 21, 2004
NONO RJOH:
That is not what I was trying to explain.
Permutations was a wrong term to begin with. I used it just because the others used this term and I did not want to confuse them.
What they meant was not permutations, but combinations. Now in a 42 number lottery there are 5,245,786 possible combinations.
What I was trying to make clear was that those 5,245,786 individual combinations of six numbers may appear in a time series in almost infinite ways. I'm not talking about the order of the numbers in the combinations themselves, but rather the order of appearance of the entire combinations over time.
pffw this is something simple, but hard to explain in my poor english.
hope this was clearer.
Yes, I did misuse the word 'permutations.' I am assuming that the number of permutations in a 6/49 lottery are shown with the calculation below, right?
The number of permutations of 6/49 = 49 x 48 x 47 x 46 x 45 x 44 = 10,068,347,520
The number of combinations is achieved by dividing the duplications, if order is not important, right? That's our 13 million number.
Of course, now that you are showing that you go back to past drawins and compare and find patterns in those relationships, you are dealing with quadrillions of possibilities, and those will be well below the means of any desktop for decades to come. So I am actually astonished that you're able to do this meaningfully in runs taking less than 24 hours.
What you're really doing is using a pseudo neural net. Neural nets go through learning processes that do what you appear to be describing. Training a neural net for 6-dimensional vectors on say 200 past drawings would indeed take up 20 hours easily. What appears to be the problem is that you don't use a neural net. You are using heuristics to approximate the behavior of a neural net, but you're doing it at prediction time, over and over again, rather than doing it once, at training time, and then making predictions from that over and over again in seconds.
From your description, I would assume that given a certain lottery, say a 6/49 with current draw history, running it for say 10 hours, you get a certain set of numbers.
Do you, if you run the same thing again, get the exact same numbers? I would say that you do. Right? That would also indicate that you're using pseudo neural net technology. A neural net, given a certain state, would yield the very same ideal results vector every time. It would be its single best bet for the next draw.
Consequently, if I were to use a neural net, I would issue at least an 8 or 10 dimensinal results vector, which would give me additional numbers to wheel with. It would be a better use of the apparent pattern.