i agree but it depends. a ranking systems by 1st and 2nd isn't the best method for sure. when i test them i prefer using a score system (using kills and casualties mainly) and having them fight against a sample of units selected to make results more stable. in some cases units are very close together while in others there is a huge difference in performance. that's why i will always be of the idea that testing is still an indicative measurement but one that can give you that extra info to enhance your game.comparison between units "who wins" when they are using very different weapons doesn't bring much more infos : it only leaves to hypothesis about why's.
Anyway man, you made me a better player with your knowledge today, i can only thank you for that.