You might want to seriously look at this: The Secretary problem.
The problem is hiring the best secretary when you have n candidates, and after you interview each one you have to immediately accept or reject them (like the stats roll).
The solution is apparently to interview n/e (e being roughly 2.718281828459) candidates, noting their stats and then rejecting them, and after that hire the next one which is better than the ones you interviewed.
So in your case, if you can do 100 rolls, total up their stats (turn them into a number) and after 100/e rolls (37 rolls in other words) then accept the next roll that is better than the first 37.
This strategy works around the problem of your having a limited number of rolls, and also means you don’t have to decide exactly what stats you want. It will give you the best stats 37% of the time. You may say, of course, but I want the best stats over 1000 rolls, but the best stats may have shown up in roll #1, and it is too late to go back and accept it.
To put it another way, even if you did 1000000 rolls, and had to stop at the last one, the last one might be quite bad, but roll 800000 was much better (and you might never achieve the stats you have specified in your script, even after a billion rolls).
|