ExpertLotto said:
i'll start thinking about some optimizing algorithm that would reduce and given set of lines while trying to cover as many triplets as possible.
When you do that, it will be a really nice to have software, especially if you add a possibility for covering 3, or 4's ,or 5's by the choice
As said I think is a good starting point splitting all starting lines into triplets, delete duplicates, and try to cover as many as possible with minimum amount of your starting lines.. in the case you wish one 3 of 6. Of course, for a wish of two 3/6 you need a little bit different but pretty similar procedure by making sure any 6 has two 3's covered, or any 5 has two 3's covered if you want two 3/5).
For covering 4/6 or 4/5 - splitting into 4's, deleting duplicates, and cover as much as possible with your lines. In case of some rest - well cover them with a combination that takes them all in.
The most important here in this procedure while using these lines is looking for an aproximately average appearance of numbers , alowing all lines equally to make a cover and not only the lines that were used first and from the top of a file.
Well, I spend a quite few days thinking about this idea, but feel absolutely free to consider and use it, and I'm sure it might work.
ExpertLotto said:
i must be a bit slow but i didn't quite get it again. 
our software of course allows working with all combinations possible. the number of all 3 even - 3 odd combinations for a 6/49 lottery is 4655200.
So, for example any 3 even and any 3 odd are drawn. Is it possible to make a cover that will have for sure at least one 3/6 (or 4/6), starting from these 4655200. It doesen't matter if that may be 170 or even 300 combinations at the end (or a several thou for 4/6).
However, since you started to consider making an algorithm, probably if not now, it will be possible soon?