The factory method is the most accurate.
(I've done dozens of engines, both ways, and compared the results. From what I've seen, Ken's product never varies more than a couple of degrees, at most. And no one can detect a couple of degrees difference on a street engine, so it is certainly accurate enough, for that application.