For the regionals predictions in DIII. 50% of runner fell within one std deviation of their expected finish, and 80% within 2 std deviations. This compares to 68% and 95% in a perfect model (normal distribution).
This round I have added simulations, which take into account a runners variance within the given nationals field. I expect this to increase accuracy.
I've worked out a few kinks for the DIII stuff but not really DI. For example, there is a chance I should weight regionals less. Let me know if you have any thoughts. There is a good chance I have the wrong top 7s for a few teams (I used whoever ran at regionals).