Talk:Fall 2015 Event/@comment-27119040-20151125161313/@comment-84.222.216.193-20151125192306

That's right, but that's assuming you have enough recorces to do "infinite" runs: in the end you will average at the expected percentage. The problem is that in these cases, the lower the percentage, the larger the variance becomes. This, coupled with finite resources and limited timeframe for the event, may results in getting nothing at all, since you simply won't be able to produce a meaningful sample size for the law of large number to take effect.