Misleading Stats vs. Misreading Stats

People Eating Ravioli.
Pookey Earning Recs.
Placer Education Reform.
Plywood Edges- Rough.
Prius Emissions Rising.

For one thing, a young player who is playing 15 mins a game, has a high TS% and a low TOV% will have a better PER than any young player who is doing normally for his age. Marresse Speights is the king of looking good at 15 mins a game and then having his PER go back down to normal levels. He’s just 1 example that springs to mind for instance.

Do you need PER to really tell you that LeBron James is a great talent? No. You don’t, or you shouldn’t. You can use plenty of metrics to find out the differences between LeBron James and Kobe Bryant. (I still take Kobe every time myself. You want a killer if you want rings. Kobe is, and LeBron ain’t. Anyways…)

I think the one thing PER has done is highlighted A) how difficult it is to come up with metrics that do really well across the board in basketball and B) that perhaps the statistical revolution won’t happen with regards to basketball because so much of the game is psychology and chemistry that numbers can’t accurately quantify.

Having said that, I look at TS%, D-Reb & O-Reb%‘s, USG and TOV% among other things. I look at things that players can consciously work on. Players tend to be efficient, or they don’t. Kobe Bryant has never had a high TS% (never once over 58% for a single season) plus a career TS% of around 55%. LeBron James has a career TS% of 56%. Obviously, on terms of TS% I’d rather have LeBron. But Is TS% the only way to measure effectiveness in scoring and value to a team? I think not.

I care about USG because USG% changes how a player’s efficiency may work. Amare Stoudemire has been very efficient (60%+ TS for most of his career in Phx) at a high USG when in the desert. Will that translate to New York now that he doesn’t have the killer setups from Nash (and the attention). It’s not that Amare’s talent didn’t help Nash, and obviously vice versa, but now there is more of a load with less talent around Amare in New York. If Amare remains very efficient and with the lesser talent, maybe it means that most of us didn’t realize how good Amare was. (Not a fan anyway, but that’s me.)

At any rate, a big reason I discount PER is that it includes a number of stats that I think have no bearing on a quality of players. Assists, blocks and steals chiefly among them. They all factors into Hollinger’s equation and how he builds PER. Obviously, other things have greater weights than others, but my problem is those things really don’t reflect a real quality of a player. Is defense gauged solely on steals or blocks? Of course not. Were Webber & Divac bad defenders because they averaged about 2 blocks a game between them? No, of course not. They were among the 3 or 4 best tandems of defensive big men when the Kings were at their best.

To recap my view on assists: Worst stat in sports. Period. There is no more subjective stat in any sport than the assist stat. Zach Harper had a great piece at Hardwood Paroxysm about how good Darren Collison’s 16 assist game really was. (And, he could have picked a number of players across the league on just about any night. I’m not sure whether that speaks to the subjectivity of the assist stat or the fact it’s so pointless because it’s so widespread.)

There is so many things that PER leaves out and so many critical things that I’ve never been able to take it seriously. Hollinger is neither stupid or clueless when it comes to the numbers thing. He’s not a big part of the APBR board for nothing at all. On the other hand, some of Hollinger’s opinions stem from PER and I think it’s a very limited tool that really doesn’t shed any light.

Any questions?