The Numbers Game

by Russ Bengtson

Today I’d like to talk a little bit about statistics. Please keep in mind that I haven’t taken a math course since an ill-advised (yet required) semester of calculus in the fall of 1989. So numbers in general are not my strong point.

But as a basketball writer (fan, junkie, what-have-you) you can’t avoid numbers these days. It all goes back to Bill James, I suppose, and his work in baseball. And now you have John Hollinger and David Berri and guys like that dissecting hoops. For years, we only relied on the basics: points, rebounds and assists per game, things like that. Shooting percentages. Wins and losses. Now? Now we have equations like player efficiency rating (PER), and something called ‘wins produced’ per 48 minutes (WP48). These numbers no doubt give a more complete picture of the game—statistically speaking. But do they help clear things up, or do they just muddy the picture further?

It seems to me that many of these ‘new’ numbers (along with some of the old ones) and the conclusions that accompany them, should be taken with roughly an ocean’s worth of salt (here’s just one reason why). Also, keep in mind that a lot of these numbers come from economists. Have you seen the economy lately? But I digress.

Berri, one of the economist authors of Wages of Wins, has written quite a bit about Allen Iverson. About how he’s far less of a player than his reputation would lead you to believe. The numbers say that, and the numbers don’t lie. He’s an inefficient scorer, turnover prone, doesn’t give you much of anything as far as rebounds (and fewer assists than you would like out of a sub-six-foot player). Hell, you don’t even need to look at the numbers to see that. Just watch him play.

Here’s the thing, though. I feel like there are things that can’t be qualified—or even quantified*—by numbers alone. Yes, numbers say that Iverson is a turnover-prone, high-volume, low-percentage scorer. But watching the game says that his speed, fearlessness and straight-up audacity allow him to get shots off when it would be all but impossible for any other player to do so. He hits shots that other players would never even get a chance to take, plays through injuries that other players never would. How do the numbers explain that? Which should we believe? Or how about this all-too familiar scenario: Iverson blows by his man on the perimeter and drives the lane. Marcus Camby’s man moves over to contest the shot, Iverson floats a layup over the top. The shot misses, but Camby—freed by Iverson’s move—is able to dunk the uncontested rebound home. Camby gets the rebound, the easy shot, and the statistical glory. Iverson gets the miss and the downgraded shooting percentage. But who made the play possible in the first place? Aren’t the numbers at least slightly misleading? How do you judge a once-in-a-lifetime player with mere numbers?

*um, if that’s even possible. Did I mention I got a C in calculus?

Maybe there should be a way to adjust statistics for guys like Iverson and Kobe Bryant, who take a disproportionate number of their team’s shots and thus are forced to take more difficult shots than, say, the three-point specialist who takes 20 fewer shots a game and only shoots when he’s wide open, or the center who just dunks. A “true” shooting percentage that better reflects shot difficulty. (Is it really fair to weight Iverson’s shooting percentage the same as Dwight Howard’s?) I don’t know—I’m no mathematician. That’s not my job. My job is to recognize basketball genius when I see it, and I say that Allen Iverson is one.

Another problem I have is with per-48-minutes stats. It’s obvious that a talented bench player will wind up with inflated numbers compared to someone who plays 48 minutes a night. Because, unless a bench player is being strictly monitored, he’ll play more minutes when he’s playing well, and less when he’s not. Stats per 48 don’t take into account fouls, or fatigue, or anything else. You’d think there would be some sort of adjustment made in regard to how many minutes the player DOES play—because isn’t that what really matters? A player like Jerome James may put up outstanding per-48 numbers, but what does that matter if he can’t play more than 15 minutes without begging to come out? Renaldo Balkman (I don’t mean to pick on the Knicks, but…) may be an extremely efficient player, but he also commits a heck of a lot of fouls. Instead of expanding numbers to per-48, how about reducing them to per-minute? They’d still be skewed in the case of guys who play 10-15 minutes a game (as those guys should never get as fatigued as guys who play a full game) but at least they wouldn’t be covering minutes that they’ll never even play. And at least you won’t have casual fans wondering why the guy with the gaudiest per-48 numbers only plays seven minutes a night.

I guess the most difficult thing about individual stats in basketball is that, of all the major sports, team is more important than anything. Five guys, playing as one, on both offense and defense. Everything fitting seamlessly into the next, no two players carrying the same load. Try another scenario: Two players play 35 minutes. One ends up with one turnover, the other six. Where’s the statistic to show that player one controlled the ball for two minutes and threw three passes, the other for 16 minutes and threw 13? Right, there’s assist-to-turnover ratio. Of course. Well, what about the point guard who makes 17 perfect passes to the exact right players that SHOULD lead to assists, but his poor-shooting teammates miss six open jumpers and four easy layups—oh, and two fail to cut when/where they should—and said point guard only winds up with five assists (and two additional turnovers that weren’t his fault)?*

*any resemblance to one of Stephon Marbury’s lines circa 2001 is purely intentional

I understand that statistics aren’t the end-all and be-all. That observation plays a large role in any actual basketball decisions, whether it be trades or draft picks. But it seems that as statistics become more ‘sophisticated,’ they’re given more and more weight. People look at something like PER or WP48, and think ‘wow, that’s complicated—obviously it means a lot.’ Sometimes, perhaps, it doesn’t mean as much as you’d think.

Please feel free to point out errors in my logic (or numbers) or say “you have no idea what you’re talking about.” Or, tell me that I’m right and that I should write a book called “Numbers Ain’t Nothin’ But Numbers.”