I have never had the opportunity to try this method but after much theory crafting and many useless pointing meetings and much statistical investigation with negative correlation between "complexity" and "time to completion" I can only think that this is the correct way to get velocity.