I can't figure out how to directly respond to Bob Ravera's thread on the One Mile Ride, but anyway:
I designed it years ago, and estimated it using geoidal elevation in a Delorme GIS product. It was almost exactly 5280 vertical as measured by that, but it was based on a National Elevation Dataset elevation model from the 90s, so there is a good chance there are better models to use now - there have certainly been surveying datu tweaks since then. If you use an uncorrected GPS, you'll get elevation as Height Above Ellipsoid, which is less precise and could be higher or lower than geoidal elevations (ellipsoids are smoother mathematical surfaces and geoids are more irregular measured surfaces of equal gravity).