With adding the 3sum into the back testing, my overall odds of wining the minimum or greater has increased and it is higher then the posted overall odds. I was using Foxpro, but with the latest windows 10 updates, it has broken my backtesting program. I am rewriting it in Java (and learning java), and then I will have to redo all the backtesting. Basically I was using different seed values for a random number generator having it create a set of numbers (with certain checks) and see what % won and calculated cost/profit if I used that specific seed value. I would feed it seed values from 0 to as high as 5 million and loop thru all the game history to calculate the results. Since illinois lottery uses a RNG, I am using a RNG to hopefully closely match theirs. I doubt I will get a 100% match, but if I did I would probably report it to them.
Basically its
for seed = 1 to bignumber:
for myrecno = 1 to totalRecordsOfLottery:
Generate prediction set of numbers meeting my requirements
See how many numbers matched and update the totals for that seed
Recalculate the overall odds if I matched the minimum needed to win any prize and update the totals for that seed
Calculate the dollar amount I won/lost and update the totals for that seed
I also started adding position match count per seed value
end:
end:
Then in a table I store all the results per seed value and I can sort by overall odds or win$ amount. A partial example of the results is in a response upthread.
I also played around lottery wheels where I would generate a larger prediction set, and sort them based on a lottery wheel scheme, then check for matches. Using lottery wheels, I found that the % didn't change much, but I haven't tested it with 3sum yet.