Story URL: http://news.medill.northwestern.edu/chicago/news.aspx?id=226405
Story Retrieval Date: 3/27/2015 5:17:25 AM CST
PART FOUR: PRO FOOTBALL FOCUS REWINDS THE TAPE
The football game is mercifully nose-diving to its conclusion. The Jaguars are close to blowing a game they’ve controlled all night. But it doesn’t really matter. These fantasy football players care about seemingly every number except wins and losses.
Back in the basketball game, the Bulls are pulling ahead by 20 points on the defending champion Heat. In the concourse, the action is more tense. The beer taps say Goose Island, Sam Adams and Shock Top. Bottles of Crown Royal whiskey line a shelf behind the bar like a pool of fantasy players waiting to be drafted.
Sure, the game is on TV. But this beats the cheap seats, where the vendors are peddling light beer.
A Matt Schaub interception is greeted with cheers and high-fives. One fan is playing against running back Ben Tate, and the interception will keep the ball out of his hands for a while. The fan pulls out his phone and launches a fantasy football app.
“Ben Tate’s projection had to be higher than 11,” he says hopefully.
Fans like him are conditioned to think in terms of expected value and win probability whether they realize it or not. Inside the arena, these are the same principles NBA teams are using to think about shot selection and lineup optimization.
With 3:45 remaining, down by four points, the Texans fail to convert a fourth-and-2 at the Jaguars’ 13 yard line. The one-week-old New York Times Fourth Down Bot, a computer program designed to do a cost-benefit analysis of fourth down strategies, soon tweets, “It didn't work out for the #Texans on that 4th and 2, but from where I'm sitting, it was a good call.” The tweet of support is likely to offer little solace to Texans’ coach Gary Kubiak, who was fired after the game.
The Times’ robot uses a model created by Brian Burke, who founded Advanced NFL Stats. His website is one of many dedicated to using math as a better way to understand football.
Another site that has gained popularity in recent years is Pro Football Focus.
“There’s been a lot of growth in the last year,” said PFF’s editor-in-chief, Rick Drummond. “It wasn’t embraced from the get-go but it didn’t take long for people to see that there was value.”
PFF is similar to STATS, in that they break down every play to the smallest level possible. But instead of high-speed cameras, PFF does it the old-fashioned way— rewinding the tape over and over.
PFF tracks two things— player participation data and player grades. The company's seasoned analysts grade the players, while a group of college students and 20-somethings keep track of who’s in the game on every snap.
More accurately, PFF seems to be where SportVU was two years ago— with some teams secretly absorbing the data and others failing to hop on board. PFF hires teams as clients. And like SportVU before the NBA put the cameras in every arena, the only teams with access to the data are the ones willing to pay for it.
PFF also has packages available for fans, who can purchase the stats on their website. Drummond said many NFL players even buy subscriptions just like regular fans. The snap counts are all objective, but if a player isn’t happy about his grade, well it’s not unheard of for PFF to hear about it.
“We offer a lot of stuff for lineman, who don’t really have stats anywhere else,” Drummond said. “They can’t go to ESPN and find stats for themselves on there. They’re people— they like looking at their own stats. Everybody does. It’s not just running backs and receivers who like looking at their numbers.”
It’s also front office members who want the data.
“There’s a big difference from one team’s analytic department to the next,” Drummond said. “I think everyone’s getting on board but they’re not all there yet.”
Drummond wouldn’t divulge whether the Texans or Jaguars are on his client list. Teams pay for both stats and client anonymity. But the New York Giants are on the list. The Wall Street Journal published a story about the Giants’ use of PFF the day before they beat the New England Patriots in the 2012 Super Bowl.
“There were phone calls that followed immediately,” Drummond recalled.
Drummond said some teams use the data in player evaluation, while others use it to study tendencies for the purpose of in-game strategy. And since more information is available for teams than what appears on the website, the teams missing out likely don’t even know what they’re missing.
Wherever there are champions of analytics, there are also detractors.
“There are always going to be people that are against it,” Kopp said. “Not even necessarily old-school people, but people who feel it takes away from the art of the sport. To me that’s not the point. The point is, just like in everything else you do, more information is better if you use it the right way.”
Almost every stat was, at some point, new. The NBA started tracking steals and blocks in 1973. The NFL didn’t keep track of sacks until 1982. So drives per game, free throw assists and opponent field goal percentage at the rim still have time to catch on.
With 3:58 left in the fourth quarter, Luol Deng hits a three-pointer to put Chicago up 17. With 1:50 remaining, he hits another three to put them up 20. There will be no Miami comeback. The Bulls will win 107-87.
Is Deng simply in the zone? Is he riding the wave of a statistical anomaly? Or is he merely shooting from the spots on the floor where empirical evidence says he’ll make the most shots?
It’s hard to say.
That’s the thing about data and analytics in sports. It doesn’t tell us what will happen, only what is most likely to happen.
Sport can be beautiful, and unless your job is to create a model that will bring your team a championship, the unpredictability can be one of its most irresistible appeals. Fans buy tickets or watch on TV because nobody knows what will happen.
But the data gives us a better idea. And more and more groups are racing to collect it.
Mitch Goldich is a sports science and technology reporter for the Medill News Service. Follow him on Twitter at @mitchgoldich.