did
The RAC option is designed around a certain type of setup and is geared toward a 4of5 match. The more you reduce
the more likely you will loose the 4of5's. There are limits to how many sets can be reduced and still maintain a 4of5.
If your setup produces or traps all 5 winning numbers the more likely RAC is going to trap at least one 4of5. RAC is
math based and does not wheel the numbers like a normal wheel. It wheels the entire line and starts with the assumption
that the winning line is somewhere within the temp file. It first has to pick the starting set at random but then uses
pure math to reduce from that point. The first run it treats each line as if it was the winning set and checks it against
all the other sets. From this information it is able to calculate how many total numbers are used in all the sets then from
this it calculates the number of sets that have none of the winning numbers. If you have 15 different numbers and each
set has 5 numbers then you can build 3 sets with no repeating numbers. Find three sets that have no repeating numbers
and here you have something that most don't consider. Lets say that the first set has none of the numbers of the other
two. Next we find another set that only has 1 number from one of the other sets then one that has two and so on and
on. If done correctly then one of the reduced sets will have a 4of5. This is why I say to use a certain type of setup. Once
this is complete then these remaining sets are optimized to reduce as much as possible. The next stage does it's thing
and so on and on until you have the best possible lines that the offer the best coverage from the temp file. There is a
limit to this process that when reached it cannot continue to reduce. Once that limit is reached the program must choose
a set at random to remove from the list in order to continue reducing. Each time ii has to do this, it a gamble but it also
uses a sort of biased approach to ensure the random selection meets a few requirements before it is removed. All this
takes millions and millions of calculations but it does very well. I have a much more advanced version but it is not finished
yet.