For all the concerns the Ravens might have about taking a first-round wide receiver in next week’s NFL draft — value, fit, the middle finger of history — they can be assured of at least one thing: He will not be slow.
At LSU’s Pro Day, Terrace Marshall Jr. ran a 4.38-second 40-yard dash. Minnesota’s Rashod Bateman and Florida’s Kadarius Toney each posted a 4.39. Mississippi’s Elijah Moore came in as quickly as 4.32. By the unofficial times measuring straight-line speed over an arbitrary distance inside a controlled environment, these are fast prospects.
But as the Ravens and 31 other teams approach an unprecedented draft, their challenge of finding a signal amid the noise of testing numbers and statistical production and historical comparisons is perhaps more daunting than ever. The coronavirus pandemic kept college scouts off the road in the fall and canceled the NFL scouting combine this winter. Big-name players opted out. Some small-school programs never played. Front-office officials flocked to Pro Days where, as ESPN draft analyst Mel Kiper Jr. quipped, “I could maybe crack 5-flat” in the 40-yard dash.
The teams that can best bring order to the chaos, that can separate game speed from track speed, may be the ones most willing to fill in the blanks of old-school scouting with new-age technologies: GPS tracking, computer vision, proprietary metrics.
“Information is everything,” Ravens coach John Harbaugh told the NFL last month in an interview tied to the league’s Big Data Bowl, an annual analytics competition that has been a breeding ground for statistical innovations in the sport. “Information is king, I guess, you might say. The opportunities that we have now to just get so much more information through technology, different programs, different areas, different places, that’s what you try to make sure that you try to take advantage of.”
Few franchises may be better positioned than the Ravens. According to ESPN, they not only entered last season with one of the NFL’s biggest analytics departments, but they were also considered, according to a survey of staffers in the field, the league’s most analytically advanced team. With their embrace of aggressive fourth-down play-calling and 2-point-conversion strategies, the Ravens have in many ways championed the most basic wishes of the sport’s quants.
In the draft, however, the Ravens’ analytical applications have been more opaque. Under general manager Eric DeCosta, the front office has remained aggressive in its pursuit of extra picks, whether through the compensatory-pick process or via trades. “It’s a luck-driven process,” DeCosta said of the draft Monday. “If you have more picks, you’re going to hit on more players.”
His first two selections last year, however, were an inside linebacker (Patrick Queen) and a running back (J.K. Dobbins) — generally two of the least valuable positions in the NFL. He didn’t draft a wide receiver until late in the third round and didn’t take an edge rusher at all — two of the sport’s most valuable positions, and now urgent draft needs in Baltimore.
In their scouting processes, the Ravens take an inclusive approach. Director of player personnel Joe Hortiz said last Monday that the team’s scouts still time prospects with a stopwatch at Pro Days, where more accurate laser timing is typically unavailable. Those hand-timed numbers make possible what Kiper called “historical-based evaluations” — comparing, say, Bateman’s 40 time to that of a previous receiver prospect.
But the Ravens also consider GPS tracking data from companies like Catapult, which uses wearable technology to measure college players’ top speed, acceleration, deceleration and more.
“You use it to marry it to what you’re seeing,” Hortiz said. “Obviously, our scouts go into schools, and they watch film, and when you put the film on, your eyes are telling you whether a player is fast or not, or whether you perceive him to be fast. But then you’re able to take that data and that information and add it to the process, and hopefully, it marries up.
“We have a way of using it all together, and our analytics guys do a great job of working their end, and our scouts, obviously, assess the speed category in their eyes. It kind of works together. Sometimes you have disconnects, and sometimes you have the marriages, but it’s great when you see it with your eyes, and then the data backs it up.”
Even if the Ravens’ draft board is shrouded in mystery, it doesn’t take Ozzie Newsome to divine what they look for, at least on offense. The Ravens were the NFL’s only team with a quarterback (Lamar Jackson), running back (J.K. Dobbins) and wide receiver (Devin Duvernay) who eclipsed 21 mph as a ball-carrier last season, according to the league’s Next Gen Stats — and wide receiver Marquise “Hollywood” Brown might prove faster than them all.
That concentration of speed is no accident. Over the past five years, among drafted players who ran the 40 at the scouting combine, the Ravens have added to their offense the second-fastest quarterback (Trace McSorley), fifth-fastest running back (Justice Hill), 17th-fastest wide receiver (Duvernay), and 18th- and 19th-fastest tight ends (Mark Andrews and Hayden Hurst, respectively). And that doesn’t even count Jackson or Dobbins, much less combine star Miles Boykin.
As the NFL gets faster, the tools to measure play speed get better. One startup, Slants, uses computer vision — a field of artificial intelligence that focuses on automating the extraction of meaningful information from digital inputs — and machine learning to generate player-tracking data.
Given game or practice footage, Slants co-founder and Maryland native Omar Ajmeri explained in an interview, Slant’s technology can identify players and derive coordinate positions on the field. From there, it’s possible to determine a running back’s velocity, chart a linebacker’s path to the ball or measure a wide receiver’s separation on go routes.
The accuracy of the company’s findings — Ajmeri said players’ estimated speeds are within half-a-mile per hour of the comparable data made available by the NFL — makes evaluations easier. With the predictive statistics that Slants compiles for interested teams, an overlooked Division III wide receiver can be compared not only to top prospects, Ajmeri said, but also NFL veterans.
“Where it really comes down to is, can we provide additional information that will be useful to your processes?” he said. “So it’s not to take away anyone’s job, it’s not to replace scouts, but it’s more about, can we give you more information than what you’ve currently been given? And can you use that information to make better decisions in the draft or free agency or whatever it is?”
There are still-greater frontiers for the NFL’s analytics field. Ajmeri said Slants hopes to integrate decades-old game film into its system, broadening the scope of possible player analogues. Dan Hatman, the director of scouting development for The Scouting Academy and a former NFL scout, said accurately gauging reaction times would be the “tip of the spear” for scouting departments.
Football’s “Moneyball” movement has already ushered in profound changes across the league. The Los Angeles Rams, convinced that their coaches, scouts and executives are better served working from home, have pulled back on draft-related travel. The importance of 40 times, Hatman said, has waned considerably. NFL Network draft analyst Daniel Jeremiah said some teams have even begun to enlist the help of their training staff in evaluating GPS tracking data: Does this prospect have a problem — and is it fixable?
“It’s really become collaborative,” said Jeremiah, a former Ravens scout. “I think the days of having your data folks off in a corner, kind of punching up some numbers and passing it off, are over. The smart teams are bringing all these people in together and really helping settle all this stuff up.”
Baltimore Ravens Insider
But even smart teams make decisions that, in hindsight, can seem anything but. Corey Krawiec, the Ravens’ manager of player evaluation and analytics, supports the team’s college scouting, pro scouting and salary cap departments with data analysis and research. His goal, he told ESPN last year, is to help the Ravens minimize their biases and make informed decisions.
If only it were that simple. With a field of approximately 200 draftable prospects this year, how exactly do you determine who plays the most like a Raven? That probably depends on whom you’re asking.
“Ultimately, we’re trying to predict the future, right?” Krawiec said. “And that’s a really, really hard thing to do.”
TV: ESPN, NFL Network