by Brett Ballantini
He was found by the Bureau of Statistics to be
one against whom there was no official complaint,
and all the reports on his conduct agree
that, in the modern sense of an old-fashioned word, he was a saint
–W.H. Auden, “The Unknown Citizen” (1940)
After contributing consistently here for a couple of months and reading for much longer than that, it seems it’s long since past time to come out of the closet.
The stats closet, that is.
Yeah, the only thing less cool than admitting you’ve just turned 40 on SLAMonline is to pump up the cause of statistical analysis in basketball. Yup yup, pocket protector in the house!
This isn’t prompted as much by reaction to the Michael Lewis piece on Shane Battier in the New York Times Magazine as it is the overall tenor of discussion on SLAMonline when it comes to numbers. It’s a bit odd that there seems to be so much fear of stats, what with all the “eff PER” and +/- chides rife on these pages. Really, are the metrics folks that close to overtaking the games?
We want so desperately to believe our eyes as the first and last evidence needed to inform our opinions and handicap our favorites in the game we so love that the idea of supplementing our own personal game notes with numbers is akin to some sort of betrayal.
Numbers can be jimmied. Every stat, including the final score, is ripe for manipulation, at least by some small measure. But in the end, that final score is the only statistic that matters.
In the game box, there are scads of seemingly unalterable statistics: points, rebounds, assists, steals, and so on. But every one of these numbers are contingent on judgment calls, haphazard timing or good and bad luck. A basket is credited to the player who appeared to score on a tip-in; a rebound can be juggled, tapped or dropped entirely; assists are routinely credited without regard to the spirit of the pass-to-score nature of the rule; and even the most blatant pickpocket earns his steals contingent on team defense.
So the notion that game stats are “pure” while secondary stats like the PERs and Efficiency Ratings and Player Wins of the stats world are somehow tainting the game and altering perceptions is ludicrous. And it makes those who shake angry fists at or shed tiny tears over basketball encyclopedias seem, well, rather short-sighted.
Advanced statistics are a sign of progress when it comes to basketball analysis, not the overmanagement or mangling of it. They are an essential part of the evolution of the game.
Think about it. The year I was born, 1969, the NBA awarded all its statistical titles to cumulative category winners, not for averages. So in ‘68-69 Oscar Robertson finished fifth in scoring with 1,955 points rather than fourth, with a 24.7 average. Dave Bing finished fifth in assists with 546 instead of fourth, averaging 7.1. Nate Thurmond finished fifth in rebounding with 1,402, instead of second, averaging 19.7.
You might think this statistical subjectivity is as relevant to today’s NBA as a .400 hitter from the 1890s getting plopped into the major leagues of today. So let’s take a more recent, and relevant, example.
Defenses were long measured by points allowed, the lower the PPG, the better. The fact that by this measure a “bad” defensive team could win a whole buncha games 120-115, or a “good” defensive team could lose just as many 90-85, leaves such a simplistic statistical measure of defense in tatters.
These days, defensive field goal percentage has overtaken PPG as the supposed best measure of a defense. And that certainly is a step forward. But still, how directly does a low defensive field goal percentage correlate to a great defense, or wins? If a club is stocked with lousy rebounders, for example, it’s of little consequence how many shots they force opponents to miss; those opponents can launch bricks with impunity, knowing they’ll have two or three or four cracks at a basket every trip down the floor.
Savvy fans have moved on to possession and pace statistics to most accurately judge teams on both sides of the ball. Defensive and offensive ratings even the hardwood for all teams by judging them per 100 possessions. A fast-paced team won’t be penalized for its cheesecloth-appearing D, while a sluggish club won’t be scored poorly for clay feet on offense.
Standardizing statistics in order to measure team efficiency per possession seems about as true to the game as the final score: maximize scoring on your possessions, squelch opponents on theirs.
For the past couple of years I’ve had some interesting exchanges with some veteran (read: older) writers, on the subject of the Denver Nuggets defense. Denver is a tricky team. The franchise still carries a tag of no-defense as a holdover from its ABA days, the run—n-stun Doug Moe Nugs of the 1980s, the thin Rockies air, laissez-faire stars like ’Melo or AI, or a combination of all of this.
Well, I’m not expecting anyone in the SLAM universe to back me, but the Nuggets under George Karl have been anything but a poor defensive team. In this season to date and his three full seasons previous, Denver has never had less than the 13th-best defensive efficiency in the NBA. Chauncey Billups has been heralded, rightfully, for all he has brought to the Mile High, but when it comes down strictly to the team’s efficiency on both sides of the ball, the Nuggets are pretty much identical to a season ago. On offense, they’re up to the 10th-best offensive efficiency (from 11th last year); defensively they are ninth in the League this year (106.4 points surrendered per 100 possessions) and 10th last year (106.3).
Last season one of my colleagues, who would consistently harp on the Nuggets and their propensity to quit on plays and play “horrible” defense, half-mockingly asked me if I actually ever watched Denver play. Heh. But calling Denver a bad defensive team (or, here’s one I love, Marcus Camby an overrated defensive player because he’s more of a weak-side help defender; perhaps if he got down on all fours and growled like Kevin Garnett, Gumby could get a little more love) wasn’t just a disservice to Denver, it simply wasn’t accurate.
Suddenly, because Billups is in town, the team’s pace has slowed to seventh in the League (from first or second in the AI years) and the antiquated points allowed per game from 107 to 100.1), and Denver has a great defense. Please.
All right, one more example before I go, so let’s look into the eye-popping world of standardized stats, an innovation fathered by Michael Goodman and beginning to gain some true traction among savvy fans. A great recent example of the power of standardized stats is in Neil Paine’s astounding column on Justin Kubatko’s Basketball-Reference, which has literally changed the way we’ve seen the game since the site’s inception five years ago. Paine suggests that Oscar Robertson’s magical triple-double season of ‘61-62 (as well as Wilt Chamberlain’s preposterous 50.4 points and 26 rebounds per) was heavily influenced by the pace of play in the game of 47 years ago. Back then, teams averaged almost 125 possessions per game; today, teams get 92.
You think a lotta tasty numbers might get gobbled up in the process of losing 33 possessions per game?
Big O is one of my favorite all-time players, and interviews. He’s dead honest about the game, with just enough cranky old-schooler sprinkled in to be a delicious quote any time, on any subject. And when Big O grouses about “giveaway” assists these days, that the assists of his day were “true” assists, no one bats an eye; hey, he’s Big O—and he’s right. But if you suggest that at the pace of today’s league, Robertson’s 30.8 points-12.5 boards-11.4 assists would be downgraded to 22-8.9-8.1, and you’re sure as hell going to serve some detention with David Stern.
Naturally, we can’t just slap a new coefficient on a season and say, boom, here are the “real” stats and, roasted, here’s your new digits, O. There are a lot of other things to take into account (not the least of which: How the hell did those fellas sprint and pound their way through possession after possession on loose floorboards strapped into Chuck Ts and not die of shin splints and related maladies?) so no one should suggest taking an eraser to the record book.
The illuminating aspect of Paine’s piece may be what it sheds light on from today’s game. The NBA’s newest triple-double troublemaker, LeBron James, would be rocking 40.1 points, 10.3 rebounds, and 10 assists if his molasses Cavs were playing at a ‘61-62 pace.
Point is, there is revelatory research going on all around us today. Pioneering possession-stats researcher (and author of the amazing Basketball on Paper) Dean Oliver works now for the Nuggets. Kubatko reinvents his site, and the way we see stats, every day, with his four factors and acknowledgement of “advanced” statistics like PER and win shares. John Hollinger at the omnipresent Big E has developed and evolved his all-encompassing Player Efficiency Rating, and you can eff it all you want, but when it comes to player prognostication, there’s not a better and more entertaining writer in the business. Goodman created EWins and has pioneered research in standardizing numbers to better understand how individual players contribute to team wins. Interesting debates and discussions go on every day at APBRmetrics (“the statistical revolution will not be televised”). And, oh yeah, basketball metrician Daryl Morey is the GM of the Houston Rockets.
There are several sites beyond B-R and APBRmetrics that will revolutionize the way you think about basketball, from 82games.com to Knickerblogger.net, and so many others. These sites and statisticians have forged a statistical renaissance that been long overdue in basketball. The movement has changed the game forever, and for the better.
The anti-stat sect wants to penalize stats that seem to minimize or skew the game, while leaving unacknowledged all the dumpster-ready “traditional” stats that can skew perceptions. You can’t have it both ways. If you pick and choose which stats to believe in, whether or not they’ve made it onto the back of your basketball cards, you’re…gulp…manipulating the game just the same.