I am looking for a series of methods that don't have to be monitored. The method I mentioned above can
be turned into a user adjustable filter which makes it much more powerful. Lets say I build a string with 25
numbers form the last however many games. It's almost a sure bet that at least one and less than five
will be in the next drawing. Since there would be 14 numbers outside the string and we could cull any 5
number combo coming from the 14 which would give us a 2002 reduction. The odds for the 5of5 coming
from this group would be 575757 / 2002 or one in 278 drawings, very safe bet. Next we can calculate the
number of lines that can be generated using the 25 lines in the string, remember no set with 5 numbers can
come from the 25 number string. There are a 53,130 lines that can be gotten from the 25 and the odds are
575757/53,130 or 1 in 10.8. This is still a good bet IMHO as 2 or 3 failures out of 20 games is not too bad
for a hands off filter. We now have a reduced the matrix by 53,130+2002= 55,132, almost 10% without
having done any analysis. The old version allowed the user to analyze the data and set a range so that
more filtering could be gotten. Lets say that our analysis suggested that we play 3 numbers from the 25
and 2 numbers from the list of 14. This would give us 2300 possible 3 number combinations from the 25
and 91 doubles from the other 14, 3200*91=209,300 or 64% reduction. What I would like to do is find
several like the above mentioned that could be combined so that no analysis is needed and still give a
good hit rate for the 5of5. The goal is a 50% overall reduction with a 3 out of 4 attempt hit rate. Calling
the first or last number odd or even can reduce by 75% but last time I tried that it was not very successful.
RL