Sports on Earth
Leodis McKelvin is a busy man. The 5-foot-10, 185-pound Troy product is in his sixth NFL season, starting at cornerback for the Buffalo Bills. He also handles punt return duties, which means he shuttles between defensive drills and special teams practice during preseason. The reps pile up, and the high work rate could lead to fatigue and injury. But the 28-year-old has Eric Ciano looking out for him.
As the Bills strength and conditioning coach, Ciano’s main objective is to keep the team healthy. That’s a thankless, nearly impossible task in the brutal, non-stop world of today’s NFL. But this season, Ciano and his staff have some help. The Bills are one of the squads experimenting with GPS sensors made by Catapult Sports, an Australian sports science and technology company. The devices track and log information like movement, speed and acceleration. They add facts and figures to observation, helping keep tabs on a potentially overworked player like McKelvin.
“We give daily reports on what a guy does and the coaches get weekly reports on what a guy does. If there’s something that really stands out, I’ll give a coach a heads up but never more than that,” Ciano said over the phone. “Our coaches understand it. They get it. They know understand if a guy is getting a lot of work between two areas that they might want to limit his reps.”
Welcome to the datafication of the NFL. Eight teams around the league including the Cowboys, Eagles, Falcons, Giants and Jaguars are using information provided by Catapult to tease out truths that, in come cases, go against conventional wisdom. It’s early -- only the Cowboys, Jags, Rams and a secret team used the sensors last season -- and sports scientists employed by the clubs have few solid conclusions (fewer still that they have shared publicly) but the trend is here: The NFL is awash in numbers.
The question is what to do with the information. Numbers for numbers’ sake help no one, and everyone involved is trying to learn what matters and what doesn’t. So far, it’s a guess and check and hope effort. “When we talked to a soccer team, a rugby team, or an Aussie rules football team, we could guide them,” Michael Regan, one of Catapult’s sports scientists, said. “American football is such a unique sporting environment that we couldn’t really help the teams. We told them to spend the first year collecting data.”
Most clubs are at the beginning of the data collection phase and are jsut starting to unlock the potential of the information they are receiving. But it’s still possible to figure out a few things. McKelvin provides a good example. Ciano says the Bills focus on Player Load, which includes factors such as change of direction, acceleration, deceleration, total distance run, max velocity and high-intensity running (how much distance a player covers while running at excess of 12, 16 and 18 mph). They use a color-coded scheme that quickly shows coaches a player’s workload, which potentially can stop a problem before it starts. “The guys we are really studying are the guys who participate a ton in practice, and who can sometimes get overused because they do so much, We really monitor those guys,” the former Georgia Tech director of player development said.
Additionally, the data can help a coaching staff maximize the efficiency of practices, which is increasingly essential as the most recent CBA limited the time players could spend on the field during camps. Instead of having separate football drills and conditioning ones, coaches can look at the numbers to ensure their players are getting fitness while practicing plays and schemes. The more familiar coaches become with the numbers -- and the longer they use it -- the more powerful it becomes.
“With our data, they can see what physical stuff they did and use that football stuff to bring about the physical game,” Regan said. “We have one team that we are helping to do a comparison of this year’s training camp to last year’s training camp. It’s been really interesting to see the feedback of how they’ve modified their camp so they can maximize what they can do under the new CBA.”
Information is power, but the teams using Catapult are missing a key metric: game data. The NFL does not allow individual squads to outfit their players with sensors. While the league will put tracking devices on some players this season, the data will be tested by the league office only. Without information from games, the usefulness is dramatically diminished. “In every other sport, the cool thing that teams do is that they measure what guys do in game and then they try to make their practices meet that demand. If you can’t measure what’s happening in games, you’re kind of guessing a little bit,” Regan said.
Ciano agreed: “We want to condition our guys for the way that they play and the sport that they play, and being able to see that data to find out how far a receiver really runs in a game, or what are speeds really in a game [would help] you prepare a guy for the demands of the sport.”
There is a vague feeling in the sports science community that the NFL will eventually allow teams to track players -- possibly sooner rather than later -- but there’s no set timetable. NFL vice president of communications Brian McCarthy simply reiterated that “clubs may not use any player tracking devices (other than the devices to be tested by the league office) during any NFL game” when I inquired about the topic. (He did admit that they “tested last season a number of technologies on a number of players in games at stadiums across the league.”)
Even without game data, however, some teams can take the next steps as they gather more practice information. Clubs are attempting to work the information into other advanced metrics. “They are trying to discover algorithms,” Regan said. “What percentage success rate does this guy have on an out route if he runs over 15 mph? What about under 15 mph? What’s our quarterback’s optimum distance to his receiver? What’s his optimum throwing time? How can we integrate the data from a tracker that tracks movement and bio mechanical data? How can we unlock some meaning in terms of performance?”
It’s a long, slow process, but one that hints at the future of the NFL. Even if 99 percent of the data is useless from an analytics perspective, it’s possible that the one percent becomes the difference between winning and losing.
Ironically, it will be a quiet battle to parse the numbers. For coaches and general managers, the lesson of “Moneyball” was to not talk about “Moneyball.” If a team finds an algorithm that works or discovers a counterintuitive fact after digging through the numbers provided by Catapult’s sensors, they aren’t exactly going to run out and tell another club. (Or a reporter. Both the Jaguars and the Giants declined to be interviewed for this story, citing a lack of time and an unfamiliarity with the technology.)
According to Regan, at least one team is already making very real progress in the area. “One thing we’ve seen is that receivers run their routes faster in practice than they do in the game. They will get out in the game and the quarterback overthrows them. Except it’s not an overthrow. The quarterback probably timed it perfectly. But it’s not a mistake,” he said, declining to identify where the game data came from. “Perfect practice makes perfect play, and these guys aren’t practicing how they play. They are practicing better than they play.”