It’s fun seeing the data I collected in beta proliferate out into the interwebz. 501 because I did 1 training and then started groups of 10. I planned on stopping at 100, but didn’t receive a 4* trainer in the first 250 or so attempts, even though it was advertised as a result. I remember being convinced that they removed 4*s until I received one.
Since I found 501 to be enough of a waste of time and the odds were sufficiently low enough to make HA7 viable when @gregschen was looking at the numbers, I didn’t do more and have not touched it in the live version. It’s possible the percentages have changed from beta, but unlikely. It’s also possible that my 250 pull drought is not representative of the overall population odds, but my guess is that it is definitely under 5% for a 4*.
Edit*: Looked at the pull sheet I had and it was a drought of 250 pulls.