Originally Posted by asgelle
Not the analyticcycling algorithm per se, but there have been papers published validating the model used there. The results show that the model is accurat to better than 2%, which is about the same as powermeters.
The greater inaccuracy is going to be getting the precise data to enter, i.e. the exact slope of the road, exact length, exact length at each gradient, etc.