My biggest pet peeve in basketball analysis is the use of advanced statistics as a valuation of players. Obviously stats can be useful when evaluating a player; but they should only be used as support, not as the crux of an argument. Why? Because stats are indicators, not proof. An example is steals. Great defenders generally average a high number of steals. Scottie Pippen, Ron Artest, Jason Kidd, and Gary Payton were all great defenders. And the stats back that up, as they are all in the top 25 in all-time steals per game. But does that mean that you can use 'steals per game' to define who's good at defense? Of course not. Because if you defined how good of a defender someone was by how many steals they got, then you'd be saying that Clyde Drexler was a better defender than Pippen, that Magic Johnson was better than Artest, that Allen Iverson was better than Kidd, or that Baron Davis is better than Payton. All of which would be laughable notions.
While Pippen or Payton may get steals because of how they play the passing lanes or having quick hands, Iverson or Davis may get steals because they gamble on defense. Getting steals doesn't make you a good defender. How you get steals is what makes you a good defender. And that's the problem with basing an argument on just stats. It doesn't provide any context.
Now theoretically, advanced stats is supposed to take care of that. But as much as they've advanced, I still think it's a flawed approach to base an argument on them. Especially PER. Nothing bugs me more than when someone says Player A is better than Player B because Player A has a higher PER. This bothers me because I think PER is an extremely limited measure of a player's performance, and I think there's one man that conclusively proves this: Dennis Rodman.
Rodman has a career PER of 14.6. That's below Shawn Bradley, Wally Szczerbiak, Drew Gooden, and a host of other guys. No offense to those guys, but Rodman's a Hall-of-Famer and they're not. The gap between him and them is astronomical. How much weight should be put into a stat that doesn't convey that? Or how about the fact that Rodman has less win shares than P.J. Brown? A HOFer contributed to less wins than a guy who never made an All-Star team? Or that a two-time Defensive Player of the year has a worse Defensive Rating than Shawn Kemp (who never even made an All-Defensive team)? How do any of these things make sense?
They don't, because advanced stats aren't good for valuing a player. They're good for determining things like how good a player is going to their left or how good a player is coming off a screen. Stats can measure aspects of a player's game, but they can't measure all of it. I think Dennis Rodman proves that. And if they don't accurately measure Dennis Rodman, what makes you think they accurately measure any player?