March Madness: A Reminder That Stats Can't Tell Us Everything

Also: a reminder that they don't try to.
Author:
Publish date:
NCAA champions in 2013. (Photo: www.adamglanzmanphotography.com/Flickr)

NCAA champions in 2013. (Photo: www.adamglanzmanphotography.com/Flickr)

One of the major takeaways from the most recent Sloan Sports Analytics Conference was that no one knows anything. Every year, hard drives grow fuller and fuller with all kinds of data, but sports scientists are making few significant discoveries. So far, there's a lot of noise and very little signal. The Moneyballification of sports, it turns out, is really difficult.

Enter the NCAA tournament. For three weekends every March, college basketball transitions from a second- or third-rate sport to one that transfixes the country. The reason, however, isn't the sport itself; it's the brackets, those 63 games people predict based on any number of factors both real and outlandish. Bracketology has become a cottage industry, one that will cost American businesses more than $130 million in lost productivity today and tomorrow alone.

When used well, data and statistics can add shades of gray to a black and white landscape. They reveal additional, sometimes unseen information and provide probabilities, but they do not give solid answers.

A new entrant into the wasting time space this year is FiveThirtyEight, the re-launched site from former New York Times stats guru Nate Silver and a cadre of hires. You know the story: he rose to prominence during the 2008 presidential campaign, first as an anonymous blogger, then as himself. His star continued to rise, then it soared during the 2012 campaign when he predicted the results in all 50 states, much to the chagrin of the pundits who wanted and needed the public to think the race was closer than it actually was. Silver's contract ran out with the Times, and he jumped to ESPN, which gave him a standalone site along the lines of Bill Simmons' Grantland.

The effort launched on Monday with an in-depth look at the NCAA tournament. As with all launches, it's a work in progress, but March Madness represents an exciting opportunity for Silver and his gang, the type of effort that let's you see what they can do and where they are going. It's perfect for them: a tournament rife with easily digestible numbers, figures, and statistics, an exercise in separating what matters from what doesn't. (It’s also a topic he covered at the Times.)

What did Silver and co. find? Well, not a whole lot, actually. Louisville (15 percent), Florida (14 percent), and Arizona (13 percent) have the best chances of winning, but no one is a convincing favorite. Throw all the data into the best statistical model out there, and you get very little. "This might be the year where the guy who just moved to your office from Estonia picks at random and does pretty well," Silver told Stuart Scott on Sportscenter.

And that's exactly why March Madness succeeds. It's right there in the title. Madness, which—at least in this case—is synonymous with uncertainty, with chaos, with fun, with that dude from Estonia who knows nothing about NCAA basketball or the grandma who picks winners based on jersey colors winning the office pool. Sports are supposed to be fun, to have crazy outcomes. They are supposed to be unpredictable. Knowing the win probability at any given moment is enlightening; watching a game in which the win probability looks like a seismometer during a powerful earthquake is fantastic.

"It will be a very exciting tournament, but not one where I'm going to look smart," Silver said near the end of his ESPN interview. "It's much tougher than a presidential election."

As the world around us grows ever more calculated, it's important to remember how little we actually know. Part of Silver's stunning success is his ability to convey this fact without sounding condescending. The thing I found most interesting about Nate Silver vs. the Pundits was how shockingly little the latter group understood about statistics (and how bullheaded they were about refusing to learn). When used well, data and statistics can add shades of gray to a black and white landscape. They reveal additional, sometimes unseen information and provide probabilities, but they do not give solid answers. Any worthwhile statistical approach comes from a point of humility. During the election, Silver was right, but he knows that he could have been wrong, too. Louisville might win the NCAA tournament but so could Duke. That's why, as they say, you play the games.

Related