03-21-2016, 06:29 PM
As someone who earns my living doing statistics, I feel your pain and look at their data collection as obviously flawed (your points, and others). The problem of course is translating what you want to a real set of values that gets the job done (explain exactly how to do it, in addition to saying the current process is wrong), and the fact that very few people/organizations are willing to use the proper methodology or able to find someone competent to do it. It just becomes too difficult to do and/or for people to understand. There probably isn't much profit in doing it right, or at least not much incentive to do it better than they do it now. What they do meets their needs for making money, it is "good enough" people see it as having value, but dumbed down enough to have broad appeal and be understood. Doing it right would involve somehow measuring coverage itself, using per-snap proportions (your suggestion), or maybe controlling for QB pressure, talent of opposition, presence of safety help, etc. What about zone coverage? The hardest part is having human eyeballs turn the player's game into good data. I could see it happening, but with a whole lot more time invested in having people watch and code the games, or something like that.
