yeeesh, all this holidays and monday's of is messing me up.. I thought it was only tuesday today..

Anywho, I posted late last time too, but I was busy looking at this new idea I've come up with.. and as Beaker said;

Beaker said:

**Keep us posted on what you find Snides. **

The new idea seems to have worked for this draw too.. I just have to figure out how to code it up so that Mandi can do all the work and test it for me over the history.. (for those that might not know Mandi, that's what I call my computer

)

The new idea would have picked the following..

2 4 5 7 14 15 16 20 23 24 22 25 26 27 28 30 31 32 37 38 39 41 42

23 numbers and would have gotten the results of having 6 winning numbers but no bonus..

I know, some of you are skeptics, and you think I am lying when I post something like this 2 hours after the draw results came out, but hopefully those of you that matter know me well enough to know that I am sincere and did come up with these numbers before knowing the draw results.

I'll give you a little bit of info about how it works..

Basically, Mandi has always picked numbers and rated them into groups, 7 groups to be specific. the rating system is actually 6 systems combined, those numbers picked by all the systems would then be in group 6, numbers picked by 5 systems are in group 5, and so on..

The most common groups to produce winners are 3 4 and 5, 4 being the most popular, so in the past i have ignored groups 0, 1, 2, and 6 which do produce the occassional winning number..

Recently I was trying to come up with a way to eliminate more numbers from each group (group 3 4 and 5 combined were usually around the 40 number mark..) So I asked Mandi to spit out some data into .csv files for me..

In these files I'm looking at both the position within that group (each group also has a rating system

) and decade hits from each group. I havn't gotten to far with position hit, but decade seems to work great.

For example, this draw had 13 numbers in group five, in rated order they are 30 38 17 1 13 25 10 3 19 6 32 40 11, from previous group five stats i decided that the 20's and 30's decades were the most likely to hit (They've both been the least recent decades to hit in this group) so that leaves only 30 38 25 and 32

By simply picking the decade I've turned 13 numbers into 4 and have retained the 1 winning number that was contained in the original 13 numbers.

so, to make a long story shorter, when you try to pick the decade for the next draw by looking at all 49 numbers it's very difficult, but if you divide your numbers into groups and try to pick decades with those groups it is quite a bit easier...

My groups always change, different sizes, different numbers per group.. I havn't tried this with static groups to see how that works, but it may be a good thing to try too...

That should give you all something to experiment with and see if you can impliment it into your systems..