There is a good chance that his defensive issues were exacerbated by playing on a young team with a coach that wasn't great. He scored efficiently yet being on a team with almost no spacing.
You're aware that stats like DRPM are specifically designed to correct for factors like those, right? Are you're aware that he's ranked #77 in DRPM among SFs? And it would be one thing if his offense was strong enough to compensate, but he's also got a negative ORPM.
What am I missing here?
DRPM corrects for his teammates, and their individual numbers. It does not correct for the coach, and the fit of teammates with each other.
Certain players are better fits with each other. Those numbers are figured out by running a regression that corrects for his teammates so that they can figure out someones worth independent of their teammates.
The problem is the game isn't played independent of your teammates. On that team Channing Frye did not have a great season, but when he was on Phoenix his RPM numbers were better, although those numbers are regressed to eliminate the effect of his teammates, he still performed considerably better by those numbers when he was in Phoenix because the situation was such that he could succeed.
Advanced stats are great when you use them to back up what you see, when you just say this guy isn't good because he has a low RPM you are being lazy. Fit, situation and coaching all matter.