Suppose we have a random number generator, and we do not know anything about the generator, including the distribution of its outputs. We are allowed to use the generator some number of times to get draws and observe the sample.
Then, from the sample(s), we want to find a number M such that we can say something like
"z% of the (future) draws from this random number generator will be less then M." (future draws meaning not including the draws we used to make sample(s))
How would I go about doing this, and are there any resources to go to?
Thanks!