Neural networksPrevious TopicNext Topic

After sprending the last three days reading what I could understand about Neural Networks, I feel there might be something here worth pursuing.   From what I can gather, the impossible part is determining which patterns to employ ( relative to the lottory, of course ).

What these networks boil down to is finding a weighted average that will be correct most of the time ( with emphasis on "most" ).  Has anyone tried to find a weighted average for the numbers?  If so, what did you discover?

With specific regard to the weighted average question; most of the time, at least one of the numbers will perform within its expected average -- often enough to be noticeable -- in retrospect. 

Spent some time reading this, the other day, http://en.wikipedia.org/wiki/Ant_colony_optimization

I think p3/p4 games act differently than p5/p5+1/p6 games. Might want to specify which games we're looking at, for a particular method.

What Does Weighted Average Mean?

An average in which each quantity to be averaged is assigned a weight. These weightings determine the relative importance of each quantity on the average. Weightings are the equivalent of having that many like items with the same value involved in the average.

=SUMPRODUCT(A2:A3,B2:B3)/SUM(B2:B3)

Sounds really interesting not to mention ALOT of trial and error but if you can figure this out and make these networks work you'll be way ahead of the game!

As for the "weighted average" of numbers this had me thinking......I was curious if anyone does or has in the past given the numbers some sort of rating(not sure if this is the same as or similar to a weighted average) in determining what numbers to play?

For example......Let's say the numbers 5.17.22.28.31 came out in Little lotto. There hasn't been a repeat in numbers in the last 3 games so I want to play at least one of these numbers.

To start I give all the numbers a +1

Now the number 5 hasn't hit in 12 games so I give it a 0 or maybe a -1 because at this point  I'm not sure what the number will do next.

The number 17 hit 3 games ago so I'll give it a +1

The number 22 hit twice in the last 7 games so I give it a + 3

You get the idea. Then I do this with all the numbers I'm considering based on the various patterns I'm looking at. Numbers would be given a + or - based on their activity(or inactivity) Of course I would decide  the specifics of  the rating system  to be used every time.

This way instead of guessing between 2 numbers I would  choose  the number with the higher rating.

I have never tried this...just something I thought about and wondered if it would be worth the extra step?

Raven62,

I have read your post about "Bayesian".

I've never heard the term so I "googled" to find out what you were talking about . Then I did some searching. After quite a bit of reading, I believe your onto something. I agree, it'll take a lot of work to implement that type of system. The formulas alone are why above my gray matter. But with that being said, written and programed properly, it could provide results... I'll keep reading...

That's why I like Bayesian Theorem, its so simple to understand. But finding rules that fit the thoery is hard work but worth it.

I have found several PDF files on line, downloaded, and read thru them.

There's one I'm really interested in that describes checking to see if a lottery "draw" is actually "random". No results were included, but they said that the formulas checked out correctly. I'm attempting to understand it (to many letters an formula markers) and see if it could be used (with tweaking) to "anticapate" next draws. But there are several PDF files on line, it takes time to read thru them all, then try to figure out if they could be used.

Let us know what you discover, keep looking, I will also...