Re: What would be the best limit to start at?
Ah, yes, I am talking about 1% of the total number of trials. The difference is of course most notable for very small values of EV, although clearly it drastically affects the number of needed trials.
Honestly, I think 1% is a pretty freakin tight boundary and most people would feel that probability holds even if they deviate more than a percent or two from expected value.
Edit: at least I'm not crazy. I spent a couple hours this afternoon idly pondering (while doing other stuff) how we got such different numbers. I arrived at the same number by 3 different methods (observation, the formula I listed here, and another method)
|