The core issue with proxy speed testing is the assumption that a lower ping always correlates with better performance. That's often false. It's just one piece of the puzzle, and a pretty small one at that. The real measure is how well those proxies perform in the sandbox - CR, margin, session...