Quote Originally Posted by wthvr View Post
this seems like the best analysis so far and therefore it is the most reliable in my eyes. other 'research' usually includes lucky streaks/high rolls where people did 5 rolls and got 2 girls etc. much lower sample size and heavily biased towards result oriented interpretation.
I would always believe high sample sizes much more, keep in mind that the 6.7% probably still are not representative as <100 rolls is still magnitudes below reliable data, but it hints towards droprates < 10% and therefore his conclusion to use kobans for 10x rolls only seems solid. (Given that he did not randomly came up with some numbers - but why should he?)
I've actually done that before, gotten 2 girls out of 5 tries. I hope it doesn't give you good rates the first few tries in an effort to real you in and make you keep playing, like real slot machines do. Anyway, after the first 2 wins (out of five tries and winning on the 2nd and 5th try) I did play several more times. The best I can remember, it was between 14 and 16 tries, but I was trying to duplicate the initial success. Well, I didn't win any more girls. Technically I guess you can say I broke even with about 18 tries in total, but you are guaranteed epic items with a 900 koban spin and not guaranteed anything with a single 90 koban spin. Even I believe it is better to save up and just do the 900 koban spin. This is the way it is designed to work. If playing single was a better option, the "900 koban, guaranteed to win a girl" option wouldn't be there.