I understand that, my point is that he is deadly when fairly open. So how does per take that into account? If he's open, he's the best shooter in basketball. If not, then he's not. I'm not sure how per encompasses that like you claim.
In real life, there are guys like that, and it has nothing to do with "high per potential" (nice try on that one, by the way). So here's a guy who if I set to -1, should be lights out, but if I set to +2, gets significantly worse than someone who who hits a lower percentage at -1 but higher at +2.
The point of all this is simply to disprove that per encompasses all that you say it does. There's simply no way it can. Which means that when you start making real-life comparisons, there are going to be holes, which there were above.