Analytics have been important to sports at least since the first coach timed players in the hundred-yard dash, but some pro leagues have embraced statistical game analysis much faster, and more effectively, than others. Lagging behind professional baseball, basketball and soccer has been football, mired in good measure by the sheer complexity of the game. That’s changing, and fast.
The NFL began capturing motion data on the field in 2014, putting nickel-sized RFID tags under some players’ shoulder pad to track and record their movements. This year, players and officials are tagged for every game in every venue, and, in some games, even the football has a sensor transmitting location data.
Matt Swensson, the NFL’s vice president of emerging products and technology, says tracking the location coordinates of players on the field allows the recording of players’ speed, how far they ran on a play, and how long they were sprinting, jogging or walking. Beyond such performance-based metrics, the league also looks at “football-based things” such as what formation a team was in, which players were on the field and how often a defense was blitzing.
The data has many uses. It’s shared with the teams to help them better understand player activity and analyze tactics, such as whether it might be better to punt in a given fourth-down situation or press forward. The data also is shared with fans – with information fed to broadcasters, to stadium Jumbotron screens and to the NFL website and the “Next Gen Stats” feature[bpm1] of Microsoft’s Xbox One NFL app – in a move to get football’s rabid fans even more engrossed in the game, Swensson says.
ZEBRA ON THE FIELD
The player tracking is provided by Zebra Technologies, whose RFID tags have been placed beneath players’ shoulder pads. Each chip transmits its location about 25 times for each double-chipped player, says Jill Stelfox, VP and general manager for location solutions at Zebra.
Broadcast at 6.3 MHz (to avoid interference on the WiFi band), the signals are collected by about 20 receivers placed under the upper decks around each game venue. The data is then sent to an on-site stadium server, where Zebra uses its proprietary MotionWorks software to match each tag to the correct player or official.
The data is transmitted from the server to the NFL’s cloud computing environment (running on Amazon’s AWS service).
From the NFL cloud, the data is shared with both fans and the teams.
A slice of the data that includes player speed, location and distance run is passed directly[bpm2] to CBS Thursday Night Football broadcasters so they can share it with their TV audiences. So far, only CBS is getting the data feed, but Swensson says plans are in motion to also do it for NBC games. That same data is moved directly to the fans via NFL.com and the NFL’s social media outlet as well as to Microsoft’s Xbox One NFL app[bpm3] . The data is also put up on arena Jumbotron screens. “At various points in the game, it will display who’s actively on field … a visual to let everybody know which running backs or wide receivers or whomever might be out there,” he says.
INSIGHT FOR THE TEAMS
While the data may be entertaining for fans, it could prove strategic for the teams. Each play is “completely evented,” says Stelfox, meaning that all data markers for a given play are recorded, including type of offense, type of defense, whether there was a huddle, all movement during the play, and the yard line where the ball was stopped. “That happens in two seconds, and that data goes to the NFL.”
The NFL runs custom-created analytics to deliver visualizations of the data[bpm4] to each team within 24 hours of the game, via a custom-built web portal. The NFL displays charts and graphs as well as tabular data to let teams really dig for insight. Many franchises have hired their own data analysts, though they’re secretive about how they’re using data to understand game tactics and individual performance.
“They can go do their own analysis if they wish to,” Swensson says, “but our goal with the website is to provide everybody a baseline, so that they have some level of understanding of what we’re seeing in the data.”
The league creates most of its analytics solutions for the teams in-house. The NFL runs a “primarily Java shop,” Swensson says, and data is stored in the open-source JSON (JavaScript Object Notification) format on a filesystem. “We also use a NoSQL database. JSON is the primary format we use for querying against the data, whether it’s in the filesystem or the NoSQL database,” says Swensson.
FUTURE FOR THE LEAGUE
Swensson describes the current analytics output as a foundation for continued expansion. He says the NFL will continue to develop metrics and features throughout this season and beyond.
“We have this ongoing list of things we want to accomplish,” Swensson says. “It’s a deep rabbit hole you can go down with this type of data.”
Beyond location and motion, other possibilities loom. Zebra’s Stelfox points to the potential to use detailed analytics to identify when a player’s performance is likely to flag late in the game, or how to improve training to prevent such dips. With that information, a coach might decide to pull a certain receiver out during the fourth quarter, or to rely on that player less in a critical moment.
“If you were to add hydration sensors and heart rate monitors, I think it could become really valuable to understanding [a player’s fitness] at any given moment,” she notes.
Swensson acknowledges the possibilities but says league leadership has not yet decided how, or whether, to incorporate such data. “We don’t currently tie into that,” he says.
For Swensson, all future uses of the data are guided by the twin goals of enabling the NFL’s 32 teams to play better football, and to enable hundreds of thousands of fans to better enjoy America’s dominant professional sport.
“Our goal on the club side is to create that visual record of the game so they have that information that can help them out,” he says. “On the media side, it’s to cater to all types of fan so they can better engage with the game.”
By Brian McDonough, from: http://www.information-management.com/news/big-data-analytics/how-the-nfls-data-operation-tracks-every-move-on-the-field-10030384-1.html?utm_campaign=big%20data%20in%20practice-dec%207%202016&utm_medium=email&utm_source=newsletter&ET=informationmgmt:e8312129:2047253a:&st=email&eid=8e5f5423e859a64488540fc441962c81