That exactly the type of test I would like to run -- except that your entries are based on a system -- and therefore your mileage with stops may not be independent of the entries you take. Hence, if I can create a random entry/exit system, with enough trials, I can smooth out the expected value. I will be sure to post my results (if I ever get it done...)
I think you would want to run this test on entries from your specific strategy instead of random entries. The reason you take trades from a system is that you believe the market is different after your entry signal than at other times - you think the expected return is higher than normal when your strategy says to go long for example. If the expected return is different than normal, then there is a good chance that some of the other return distribution characteristics that affect the stop performance (volatility, skew, autocorrelation, etc) will be different too.
Here are twenty five such plots, at progressively greater magnification. Convergence is slow: a million hands gets you two, but not three, digits of accuracy.