As Ravens coach John Harbaugh joked Monday, analytics have been around since his dad, Jack, was coaching. Back in those days, they amounted to figuring out what plays weren’t viable on third down.
“There’s always been analytics in football,” the younger Harbaugh said. In 2021, however, there’s “just a much more advanced version of what’s been going on for many years in football.”
The scrutiny is greater, too. The Ravens (8-6) enter Sunday’s showdown with the AFC North-leading Cincinnati Bengals (8-6) on a three-game losing streak defined as much by its narrow margins of defeat (a combined four points) as by the analytical reckonings left in each game’s wake.
In a 20-19 loss to the Steelers, the Ravens went for a go-ahead 2-point conversion with 12 seconds left, determined not to let Pittsburgh quarterback Ben Roethlisberger take advantage of a depleted secondary. In a 24-22 loss to the Browns, the Ravens went for a 2-point conversion after they’d cut Cleveland’s fourth-quarter lead to nine points, rather than wait for another touchdown to make up the ground.
And in Sunday’s 31-30 loss to the NFC-leading Packers, the injury- and coronavirus-ravaged Ravens went for a last-minute 2-point conversion, hoping one last go-ahead score might be enough to hold off Green Bay and star quarterback Aaron Rodgers. All three plays failed. All three stoked conversations across the sport about the use and abuse of analytics.
If the field can seem inscrutable, that’s because much of football is. No team has the same personnel, coaches or analytical models. But as the discourse reaches new decibel levels amid Los Angeles Chargers coach Brandon Staley’s bold fourth-down play-calling and Harbaugh’s stretch of unsuccessful 2-point attempts, it’s important to separate fact from fiction.
Fact: The Ravens are among the NFL’s most analytically inclined teams.
Just look at the results of ESPN’s October survey of NFL analytics staffers: The Ravens were regarded as having the league’s second-most analytically advanced team, the second-highest level of “analytics work” and the second-most incorporation of analytics into their decision-making — behind only the Browns in all three categories. (Not surprisingly, the Ravens are also believed to have one of the NFL’s bigger analytics staffs.)
In practice, the most mainstream influence of analytics is on fourth-down play-calling. According to Football Outsiders, Harbaugh “set an all-time record for fourth-down aggressiveness” in 2019, going for it on 23% of what the analytics website called qualifying fourth downs. The Ravens led the NFL that season in fourth-down success, converting 17 of 24 opportunities overall.
In 2020, Harbaugh was the fourth-most aggressive coach on fourth down, according to Football Outsiders’ metrics, going for it on 15.3% of qualifying opportunities. The Ravens’ conversion rate fell to 63.2%, still good for eighth in the NFL.
This year, the Ravens already have as many fourth-down attempts as they did last year (21), but with improved efficiency (66.7% conversion rate, No. 2 in the NFL). Some of their fourth-down successes are season highlights: Jackson’s 2-yard run to hold off the Kansas City Chiefs in Week 2, a 1-yard touchdown pass to fullback Patrick Ricard in a Week 9 comeback victory over the Minnesota Vikings, a 30-yard completion from Huntley to wide receiver Rashod Bateman late in the near-rally against the Browns two weeks ago.
Fiction: The Ravens’ decision-making process Sunday aligned with conventional analytical wisdom.
Harbaugh said his decision to go for two Sunday was based more on feeling than on numbers — “mostly gut,” as he put it, because “the numbers aren’t perfect.”
In this case, the numbers weren’t clear-cut, either. ESPN’s model indicated that an extra-point attempt would’ve given the Ravens only a slightly better win probability. EdjSports’ model also favored a Justin Tucker kick, but not decisively so.
The most important variable in Harbaugh’s calculation was time. In going for a 2-point conversion with 42 seconds left, he seemed to acknowledge that the Ravens would have a better chance of stopping the Packers once than they would for possibly a second time, in overtime.
In a Week 3 win over the San Francisco 49ers, Rodgers had needed just 37 seconds — and no timeouts — to lead Green Bay on a 42-yard drive capped by a last-second field goal. On Sunday, the Packers had more time and a timeout available. The Ravens, meanwhile, without four of their top six cornerbacks and their two starting safeties, were allowing 8.6 yards per pass attempt.
“Forty-two seconds was a little bit more time than you really want to go for two there; I would rather have it be around 20 seconds, but decided to do it there at the end,” Harbaugh said Monday. “If there would have been any more time than that, we would have certainly kicked it.”
Harbaugh’s more interesting decision came on the Ravens’ previous touchdown. When Huntley scrambled in from 3 yards out with 4:47 remaining to cut the Packers’ lead to 31-23, conventional analytical wisdom suggested that Harbaugh would go for two. With a theoretically 50-50 chance of converting, the Ravens would be well-positioned to either take the lead after a second touchdown or tie the game with another 2-point try in its closing moments.
Instead, Harbaugh sent Tucker out for an extra-point attempt, which he converted. Asked about the decision Monday, Harbaugh referenced the Ravens’ loss in Cleveland the previous week, when they went for two when trailing by nine and failed.
“I wasn’t quite as sure it was going to be a two-touchdown situation in this game,” he said. “There could have been another score involved, so I just wanted to wait and see how that played out. That’s why I didn’t go for it on the first score. I wanted to wait for the second score. Plus, you never know how much time you’re going to have left. So, if it had been any more time, we would have definitely put it into overtime or attempted to put it into overtime and tried to get the stop.”
Fact: The Ravens have struggled on 2-point conversions.
Every year, 2-point attempts amount to nearly coin-flip propositions in the NFL. That’s been the case through three-plus months this season: Entering Week 16, 129 have been attempted, and 63 converted (48.8%).
The Ravens’ production, however, has been subpar. After their failed conversion Sunday, they are 2-for-8 on 2-point tries this season. Among teams with at least five attempts this season, their success rate ranks ahead of only the New Orleans Saints (0-for-5) and Vikings (1-for-7).
The Ravens have fared better in previous years. Last season, they converted both of their 2-point conversions. In 2019, they went 0-for-3 in a September loss to the Chiefs, then 2-for-2 over the rest of the season.
While offensive coordinator Greg Roman has been criticized for his play design on 2-point tries — Steelers star outside linebacker T.J. Watt was left unblocked in Week 13, and Huntley’s rollout to tight end Mark Andrews’ side Sunday eliminated half a field of potential targets — the more salient critique might be of his play choice.
A 2018 study found that, over the previous seven seasons, teams passed the ball almost four times as often as they ran it on 2-point attempts, even while the conversion rate on runs (63.6%) far exceeded the rate on passes (44.9%). This season, despite J.K. Dobbins running in both of the team’s 2020 conversion attempts, the Ravens have passed on all but one of their eight 2-point tries. (Jackson was stopped short on a designed quarterback run in a Week 5 comeback win over the Indianapolis Colts.)
Fiction: Analytics aren’t helpful in the long run.
Analytics are a tool, not an orthodoxy. They help teams make more informed in-game decisions. They help shape offseason team-building strategies. They can have applications as important as fourth-down aggressiveness and as overlooked as special teams value. (Ravens punter Sam Koch told Nebraska’s Lincoln Journal Star this summer that, according to analytics, punts to the 13- and 18-yard line “might be the difference” between a 13% chance of scoring and 28% chance of scoring, respectively, for the opponent.)
Every modern sports team uses some kind of analytics, and even franchises as progressive as the Ravens can embrace old-school strategies. In 2019, when the Ravens had the NFL’s most efficient offense, they had football research coach Daniel Stern sitting next to Roman, dispensing information about in-game situations. They also rolled over defenses not by throwing the ball, as analytics would have almost all offenses do, but by leading the league in rushing.
Analytics didn’t doom the Ravens on Sunday, either. Well before the Ravens scored the last-minute touchdown that drew them within a point of the Packers, Harbaugh seemed conflicted. There was what the numbers told him. There was what his gut told him. And there was what his players told him.
According to Huntley, Harbaugh told him that the Ravens, then trailing 31-17, would score two touchdowns and go for a 2-point conversion on one. But in Harbaugh’s sideline conversations Sunday, footage of which the team released Monday, he still seemed uncertain about what to do after Huntley’s would-be tying score.
Harbaugh asked a handful of offensive players, including Huntley, “What do you want to do?” He added: “We can go into overtime here or we can go for it.” There was more to consider than just data on a spreadsheet. When one player told him to go for the win, Harbaugh agreed. The offense went back out onto the field, and Tucker stayed on the sideline.
“I’ve thought about it all night,” Harbaugh said Monday. “There are two choices, and they’re both viable. Either one can turn out right. Either one can turn out wrong. It’s basically 50-50. We talked about a lot, [and] we decided to go for it. It didn’t work out. I know half the people are going to say we should’ve kicked it. I get it; they can certainly criticize me for it. I’m OK with that; I criticize myself for it. So I understand it. That’s just the way it goes. …
“I think it’s a little bit like decisions in life. Maybe it would have turned out differently, maybe it would’ve turned out the same. There’s no guarantee we would’ve gone down the field and won in overtime, either. So that’s where we’re at. We’ll do the best we can going forward with those kinds of decisions.”
Sunday, 1 p.m.
TV: Chs. 13, 9 Radio: 97.9 FM, 1090 AM
Line: Bengals by 2 ½