The Ravens had the best regular-season record in football this season thanks to their embrace of innovative, data-driven on-field strategies.
The Baltimore Ravens are not in this year’s Super Bowl, which will feature the Kansas City Chiefs and the San Francisco 49ers. Though they had the best regular-season record of any team in the NFL this season (14-2) and their starting quarterback, Lamar Jackson, was unanimously voted league MVP, they lost in the Divisional round of the playoffs to the Tennessee Titans, whose star running back, Derrick Henry, was on a historic tear. But Baltimore made some history of its own this year by kicking off the NFL’s analytics revolution.
During NFL games, analysts such as the Ravens’ Daniel Stern, a 25-year-old Yale cognitive-science student, now sit in the coaching booth and dictate win probabilities to on-field coaches based on internal team analysis of what types of plays can deliver the highest chances of ultimately scoring touchdowns and winning games.
The Ravens’ elevation of analytics under Head Coach John Harbaugh this year contributed to their having some of the highest single-season third- and fourth-down conversion rates in NFL history, and encouraged them to “go for it” on fourth down at one of the high rates in the league, a strategy analytics gurus and economists have long suggested makes mathematical sense. Over the past few seasons, teams such as the Ravens and the Philadelphia Eagles have consistently been near the top of the league in fourth-down conversion attempts thanks to head coaches who have been supportive of such aggressive, win-probability-maximizing strategies, which for most of the league’s history had been shunned by coaches averse to risk.
Other aggressive analytics-based tactics being implemented include passing more often on first down, as passing in that situation will net at least five yards 47 percent of the time while running yields five yards or more around only 32.8 percent of the time. The Ravens and Eagles also are among the teams that most frequently go for two-point conversions after a touchdown, another strategy recommended by football-analytics professionals since a 2015 NFL rule change that moved extra-point tries to the 15-yard line from the two-yard line, making them slightly more difficult to convert and slightly less valuable than two-point conversions.
These innovative strategies weren’t born in a vacuum; they’ve grown in tandem with the availability of special new data taken from the RFID chips that have been implanted in players’ shoulder pads and the football itself for the past several seasons. In 2019, for the first time, the NFL even held a Big Data Bowl, which allowed outsiders — mostly 20-something statisticians — to compete in analyzing some of this new “Next Gen Stats” tracking data. Some of those who impressed have even been hired by savvy NFL front offices, including Baltimore’s.
The applications of the new data are nearly limitless. Already, teams are using machine learning to classify receiver routes in a system based on that used by legendary coach Bill Walsh. Once that’s done, analysts can assess what types of routes might be most effective against various defensive schemes, potentially giving their head coaches a big in-game advantage when deciding what plays to call.
The NFL RFID data isn’t perfect. It still might not pick up a juke or a stutter-step. But it has been supplemented by other imperfect-yet-groundbreaking data from Pro Football Focus (PFF) and other private firms. Pro Football Focus produces granular player rankings that record statistics such as receiver drops that aren’t tracked by the league itself. PFF has also recently come up with its own “wins above replacement” (WAR) measure, which it hopes will be football’s answer to Bill James’s highly popular formula for quantifying the production of baseball players. The firm was recently acquired by Cris Collinsworth, the color commentator for NBC’s Sunday Night Football telecasts, which feature PFF player stats. All 32 NFL teams now also subscribe to its data, which arrives in team inboxes on the morning after each game day.
But back, for a second, to the Ravens. Before Baltimore, some other NFL teams tried to embrace modern analytics to varying degrees of success. In 2016, the Cleveland Browns hired Paul DePodesta — a former assistant to Oakland Athletics general manager Billy Beane, who famously helped usher in baseball’s analytics revolution as chronicled in Michael Lewis’s 2003 bestseller Moneyball — to serve as the chief strategy officer in its front office. Under DePodesta, the Browns adopted a strategy taken from a paper written by Nobel Prize-winning economist Richard Thaler, trading down in the NFL draft in exchange for as many picks as possible, both in the present and the future. DePodesta’s reign saw the team assemble a promising young roster, but the Browns’ coaching staff has been much less receptive to on-field analytics than the Ravens’, and the wins have yet to follow.
Of course, an analytics skeptic could point out that the Ravens will head into the offseason without the Super Bowl they craved, much like Beane’s Oakland A’s have never won a World Series. But like Beane’s A’s, the Ravens deserve credit for touching off a sea change in the way their league’s teams process and use data to gain an advantage. The analytics revolution Baltimore has started is here, and its impact on the sport of football cannot be overstated.