Sports News Without Access, Favor, Or Discretion

"Sports Analytics" Is Bullshit Now

The 9th annual MIT Sloan Sports Analytics conference takes place this weekend, and you've likely noticed all the coinciding and ongoing coverage of the analytical side of sports. Some of these stories are illuminating explanations of what's going on at the vanguard of sports analysis; others bitch, correctly, about how the very coolest sports stats will almost certainly never be made public; and some great portion are just bullshit. Let's talk about the bullshit.

Consider this list from ESPN The Magazine's analytics issue, which purports to rank the teams that put analytics to the best use, and the fact that the Philadelphia 76ers and Houston Astros occupy the top two places, ahead of teams like the Houston Rockets, Dallas Mavericks, and San Antonio Spurs, or the Boston Red Sox, Tampa Bay Rays, and St. Louis Cardinals. Partly this is just a strangely compiled list in a strangely compiled magazine package about analytics, but even so, it reveals a dim sort of inefficiency that certain teams have learned to exploit in the media.


Here's a sample passage from ESPN's analytics bukkake:

Analytics say signing a 35-year-old coming off an Achilles tear to a two-year, $48.5 million extension is always a bad idea—even if he's Kobe. LA should have listened: He's been worth minus-0.2 win shares since last season and is back modeling suits courtside with a torn right rotator cuff.

"Analytics say." Holy shit! Analytics say don't sign a legless mummy. Citing this as proof of the usefulness of sophisticated quantitative strategies—they produce information that could be gleaned from other, more traditional methods, like science, medicine, common sense, or the conventional wisdom of a bunch of cavemen not letting that one other caveman on their team because he is old and has no legs—is the near-literal definition of damnation by faint praise. There might be something to, say, citing modeling that landed on a -0.2 win shares figure ahead of time (though there might not be), especially if that model illuminated something about how different injuries affect different sorts of players. As retrodictive filler stripped of any meaning by necessity, though, analytics is posed as a sort of truism engine, a mechanism for coming to the most obvious possible conclusion.

The ESPN package is filled with nuggets like this, allusions to popular stats-ish and more or less agreed-upon theories like the Curse of 370 or the abiding value of draft picks, sliced thin enough to be digestible by an uninitiated sports fan. This isn't at all how the teams themselves use analytics, but it's necessary for covering advanced stats for a broad audience. Unfortunately, it also has the collateral effect of flattening all analytics into the same rarified tone and station, washing out any qualitative distinctions and for all intents endowing analytics with magical powers. That makes conversation annoying when the armchair quants overthink a simple problem circuitously enough to crash into the ass-end of the talk radio rearguard, or manage to read a counter to the counterintuitive reactionaries as anti-intellectualism rather than critique. But more importantly, it makes talk about teams' "commitment to analytics" disingenuous by definition if the examination of that commitment ends at How much are they trying? without touching upon how much they're succeeding.


These are increasingly crucial distinctions in sports, where the information available to teams has basically settled at an industry-wide tie. Just a few years ago, the best data was brokered by independent agencies that would sell it to teams individually. Now, the NBA subsidizes league access to SportVU player-tracking, while MLBAM is rolling out its in-house fielding-tracking system, giving teams access to more data than most know what to do with. That means that there's less and less low-hanging fruit for guys like Billy Beane, who were early to the notion of, well, simply bothering to look.


This means that beyond the basic entry point of actually hiring some asshole with a background in statistics and a knowledge of your sport to do the work, having a "focus" on analytics can't possibly hold any meaning. It's like saying you're going to emphasize coaching, or really key in on scouting this year. That was good progress when this was a blind spot big enough to take in entire sports, but at this point lauding teams for saying they're really very pro-math basically comes off like the local paper blowing the mayor for saying he's going to focus on budget control. Analytics isn't revolutionary; analytics has already won, and is a rote component of nearly every organization. The hard work of gathering the observable data has largely been taken care of—by MLBAM and SportVU and before that Synergy and scouting and charting departments and other traditional arms of teams that are being augmented, not displaced, by analytics—and it's just teams' job to catch the spreadsheets and code when they fall off the back of the truck. What's important now is what you do with the information once you have it, and how well you do it. Magical endowment or not, you need the right stats wizard to cast the balrog back into Khazad-dûm or crack the playoffs in the Western Conference.

To that end, yes, of course, some teams invest more in analytical personnel, and some do a better job of integrating that into strategy, tactics, and method on the field or court. But try naming some and why, specifically, they're successful. You'll get as far as an accommodating owner or infrastructure, like the Mavericks or Rockets enjoy, or the efficiency of scale the Red Sox or New York Yankees can sometimes employ in turning small-market tactics loose with big-market ruthlessness (and margin for error), but not much further. There's a reason for that: the actual work of transcoding data to usable intelligence involves the most closely-kept industry secrets in sports, to the point that, as I heard it from some of the private analytics crowd, multiple NBA teams were furious last year when Kirk Goldsberry and the Harvard group made public their models for parsing SportVU data in a useful way. As a rule, no one shows their work; no one shows proofs or models or submits to peer review that extends beyond the trade deadline. Yes, it's probable that the Rockets really do have some great quants working in-house—who, like the best quants at your local Evil Finance Operation, are likely kept chained in a dark room being fed graphing paper so competitors never learn their names—but that probability rests in the team's proven record of turning observation into philosophy into decisions into on-court success. The irony is that for all the complications added to sports organizations over the last several years, because of those complications and their secrecy, simple old wins are just about the only honest metric available to the public by which to judge whether or not a team's analytics department has done a good job.


That doesn't mean no one tries to prove otherwise. In the absence of any way to check everyone's homework for bullshit, it's very easy to just praise the teams who seem most vociferous about abiding by the maxims of analytics as good-faith practitioners. Draft picks and youth are valuable; roster flexibility is key; defense is cheap, and maybe the new (or now just recent) inefficiency—all the little truisms distilled and disseminated in that ESPN analytics package are basically just talking points when coming from teams that promise they've done the math even if you can't see it working yet.


As such, the analytics story right now isn't just about a standard-issue business division being imparted divine powers by the media and the people who benefit from such coverage; it's about right-minded and unverifiable claims being attached to all manner of disasters as a way of reassuring fans that they shouldn't worry, that there is a plan, and that the plan is based on smart-people things. How vile or moronically incentivized the plan might be on the way to its goals tends not to matter as long as you can sell the future endpoint, which is usually a shortcut to a championship, or more likely a dynasty, probably. (We've beaten these war drums before, but it's a special kind of cynic that reads flushing multiple years' worth of games down the shitter, a strikingly anti-fan and anti-sport decision, as a revelation in modern sports strategy.) In any case—and as always, without getting into the specifics, because who wants to read about the sausage being made?—the team is positioned for the future, despite whatever we just did or whatever you might think you're seeing now.


There is a name for an operation that works like this. It's called a branding campaign. A scandalized company looking to scam the public into believing that the near-future truth will reflect the opposite of the recent past will re-brand into its idealized self, regardless of its relationship to reality. Coca-Cola is healthy now. McDonald's serves gourmet foodstuffs. BP is a leader in sustainability. Gawker isn't written exclusively for the same few dozen Park Slope millennials responsible for 90 percent of the internet's Lena Dunham thinkpieces, most of whom write or have written for Gawker. And the 76ers and Astros, two of the most wretched teams in professional sports, are implementing the very best, state-of-the-art methods of ensuring a successful future for their clubs and serving the interests of their fans.

This can work, as far as it goes! There's a reason why these teams are judged more praiseworthy than rivals that have used intensive data analysis as a crucial foundational element of actual championship runs. Their public relations campaigns have been effective. The redeeming thing about sports, though, is how brutally the results tell. Even if the worst-case scenario turns out and this Sixers re-brand ends up just being Sam Hinkie cynically fucking the last scraps of faith and good will out of Philadelphia, the wheel has to keep turning. Coke can stay selling its sugar water, and BP can stay pulling oil out of the ground, but the 76ers, at some point, will be forced to emerge from their self-imposed state of quantum undefinition and actually try to play some fucking basketball, at which point they and their analytics will be able to be judged, probably by more analytics, because all analytics really does is observe what has happened and predict what might. Which is to say that it's about reality, not blood magic and not faith. It's the description of observable reality. It doesn't care whether or not you believe in it.


There's a panel at this year's Sloan conference posing the tired old question of whether or not analytics is taking the joy out of sports. Sitting in the audience at a networking event sponsored by the media conglomerate whose magazine highlighted a basketball team creating bold new strategies to keep open the window during which it can continue to be awful and ranked the Astros ahead of the Cardinals and Rays in analytic might, an alternative question might occur to some: To what degree has analytics taken the sports out of sports analytics?

Art by Jim Cooke

Share This Story