<
>

Bad Use of Statistics is Killing Anderson Varejao

Put together whatever list of statistical experts you want. Dean Oliver, John Hollinger, Daryl Morey, Dan Rosenbaum, David Berri, Justin Kubatko, Wayne Winston, Jeff Sagarin, Jeff Ma, Kevin Pelton ... there are a lot of bright people out there working on the best way to describe basketball in numbers.

There is a lot they do not agree on.

But I'll stick my neck out a little and boldly claim that one thing they all agree on is that just looking at points and rebounds per game is a pretty terrible way to judge a player's value.

Yet that's totally standard operating procedure in basketball.

Talk about Anderson Varejao holding out for big money, and people will tell you he's crazy because he averaged six points and six rebounds per game.

They say that like it's case closed.

I have heard this from coaches, from journalists, from all kinds of people. Is it part of the actual debate between the team and the relevant agents? I have to suspect it is.

Even players are on board. Case in point, here's Kendrick Perkins (in the third part of this three-part interview series on PerkisaBeast):

I mean, Varejao, where's he getting $60 million at? Al getting 71 you could say he's worth it. He's gonna consistently give you a double-double every night - A high double-double. So he's gonna be a 20 and 10 guy every night. So every night you could depend on Al to give you 20 points and 10 rebounds, you can't do that with Varejao. You can't depend on him to do that.

Meanwhile, as Perkins points out, Minnesota's signing of Al Jefferson (for more than Varejao is reportedly asking for, but less than the maximum) is widely lauded as a great move. And when it's lauded, people cite the fact that he's young and getting better -- true enough -- but they also cite his scoring and rebounding numbers from last season: 16 and 11.

16 and 11, the conventional wisdom goes, is valuable. Six and six, they say, is not. And it seems like a no-brainer.

But of course everyone knows that assessment is lacking. It's lacking passing. It's lacking defense. It's lacking how well the guy sets a pick, or how many loose balls he collects. It's lacking field goal percentage, decision making, shot blocking, closing out the shooter, fouling, turnovers, and a million other things. In fact, it's only measuring some of the things things a player does within a second or two of some of the times someone shoots. The rest of the time the clock is ticking -- points and rebounds measure nothing.

Can all that other stuff add up to be more important than points and rebounds? All those statistical experts are finding that, increasingly, the answer is yes.

One of the most sophisticated ways to measure all that, while including defensive contributions, while adjusting for what a players teammates are doing at the same time, is the adjusted plus/minus I talked about the other day.

In 2006-2007 rankings of adjusted plus/minus, among players who played a decent number of minutes, Anderson Varejao was the 22nd best player in the NBA. As in, an All-Star. He ranks ahead of Shawn Marion, Tracy McGrady, Ray Allen, Dwight Howard, Allen Iverson, and countless others.

Jefferson, with his gaudy old-school statistics, was 128th, behind guys like Charlie Bell and Luther Head.

Several other sophisticated statistical analyses similarly rate Varejao as an elite NBA player.

Is this case closed? Does Varejao deserve more money than Jefferson? Should salaries just be set to the adjusted plus/minus list? I don't think so. I have argued many times that new breed basketball statistics are still embryonic. There is much more to come. And in the meantime, factors like old-fashioned scouting and basketball know-how need to play a massive role.

(For another thing, Jefferson is a course to get way better, we all assume.)

But to the extent that numbers are used, we really should attempt to understand and use the best numbers available. And while we might not agree on what those numbers are exactly, we should, by now, agree it's not friggin' points and rebounds.