Besides the obvious flaw in that it calculates jabs and power punchers without any mention of the power of the punchers and the impact they had, it is also wrong at times with the total number of clean punchers landed that it says did. It is still fairly accurate though, or so i'm lead to believe. It would be interesting if the error margins (+ or - X amount of punchers) of CompuBox were released.
It's far more accurate than any singular opinion of anyone. I'm sure if you were to watch an entire fight on super slo mo replay you could find mistakes they have made but neither they or the judges have that luxury. And that's what it's really all about to me, a completely independent back up system to keep the judges honest, something that says what the judges saw and scored may or may not be true. If you look and think it about that way you will see that the compubox system almost always confirms that what the judges saw could be true. I find that the judging is always talked about as being corrupt but, it's in fact the rest of the politics that are far more corrupt. Like who gets p4p status, mandatories, title shots etc. ..judging not so much, and not nearly as much as people think as it's far more difficult to be when you have nothing but witnesses watching a fight.
It's accurate enough. Maybe you can count 50 punches thrown in a round, and the guy doing the punch stats counts 48. This would be 96% accurate.
I just love that the same idiots who cited CompuBox constantly after Pacquiao-Marquez 3 suddenly have concerns about its accuracy. Anyway its only useful as a co**** measure of relative workrate.
Comp-U-Box is just two guys sitting there clicking a button. They do the best they can, but they are human.
If you could get as many people on here to count the punches thrown in just a single round I would say you would get a fairly wide swing in punches counted so to leave ir up to just two guys leaves a lot of room for error.