Ok Joe, let me know.
I´m starting from the assumption that a future draw won´t have the same "form" as the previous one. So, it won´t manage to fulfill to all the picks the same rule of transformation. As i told you may be that one, two, (and sometimes three, but more rarely) numbers are picked. All the combinations with more than this can be discarded. If you furnish the "programming apparatus" you will undoubtely see lot´s of possible transformations/permutations (though it doesn´t have to be a permutation but instead a movement from one number to the other, i.e, in some transformations 1 may be transformed in 3 and 3 in 1, but in other "configurations 1 may turn 3 and 3 in 5) to be applyed. If you gave us this instrument, then looking at the panel you can study movements from one number to another, aplly that rule to all numbers picked in the previous draw, do the transformation, and apply a file filter as refered before. Maybe the difficulty is in understanding how many transformations can be done. I assure i´ve at least 50, so this will give a file with 50 lines of "transformed numbers" from the last draw, that we can use to filter all the combinations on the package that have 3 or more equal/coincident numbers in them.
My difficulty is that doing this by hand i have to apply this transformations to the last draw 50 times and then have to create a file filter with 50 lines with 5 numbers to use in E. L. And i have to do this every week because as you understanded this is a filter that depends on the previous draw (it can be apllied to all previous draw, but in this case we should filter only combinations that are coincident only at 4 and 5 numbers).
Hope this can shed some light in the idea i presented.
Just let me know whatever you need to know better to develop this tool, and if it is possible. Though i haven´t seen all posts in E.L i think this is an unexplored possibility, that has some strenght in it. If Stan is listening maybe he too can have some insight in here.
Thanks once again
Gats.