Now that we're more than halfway through the 2020 season, NFL Media researcher Joseph Ferraiola takes a fresh look at one of the bigger analytics storylines of the past offseason.
Mike McCarthy's first season as head coach of the Dallas Cowboys has not gone according to plan.
No one expected McCarthy, brought in to help Dak Prescott compete for a Super Bowl, to be sitting at 2-7 in November, with Prescott on injured reserve after suffering a season-ending ankle injury in Week 5. Surely not even McCarthy himself expected to be coaching QBs like Ben DiNucci or Garrett Gilbert, or for Dallas' defense to rank as one of the worst in the NFL after finishing ninth-best in 2019. That's to say nothing of the other bumps in the road he's encountered.
However, we can still see if McCarthy has followed through on one of the key promises he made even before landing the Cowboys job in January: to incorporate analytics into the decision-making process and be more creative in his play designs on offense.
McCarthy's 13-year tenure with the Green Bay Packers came to a lackluster end in 2018, and the analytics vision he developed during his one-year hiatus from the league was one way to prove to potential employers that he's rejuvenated his coaching style to fit the modern NFL. Now we're nine games into McCarthy's first season back -- and it's time to see if he kept his word.
Note: All figures and ranks are updated through Week 9 of the 2020 season, representing the portion of the season for which Dallas, which had a Week 10 bye, has been active.
Basic analytic theory
Passing plays are the foundation upon which basic football analytics theory is built. Expected Points Added (EPA) is a metric derived from Next Gen Stats' Expected Points model, which measures how each play potentially affects the score of the game relative to the situation. And we know that since 2019, passes (excluding spikes and throwaways) are worth +0.20 EPA per play, compared to rushes, which are worth -0.04 EPA per play.
Whether intentional or not, McCarthy has always been analytics-friendly in his run-pass ratio. In the head coach's final five seasons with Green Bay, McCarthy's Packers led the NFL in pass-play percentage on first and second downs, also known as early downs (59.9%, per Pro Football Focus). For context, the Cowboys were last in pass-play percentage on early downs in that same time frame (49.5%).
McCarthy has certainly had a positive influence on the Cowboys' early-down run-pass ratio thus far. The Cowboys have passed on 62.1 percent of early downs, the sixth-highest rate in the NFL; compare that to 2019, when Dallas passed on just 54.8 percent of early downs, the 20th-highest rate in the league. The Cowboys have also slightly increased their pass-play percentage on second-and-long (7-plus yards to go), from 78.1 percent in 2019 (fifth-highest in the NFL) to 79.5 percent this season (fourth-highest in the NFL).
However, since McCarthy took over, the Cowboys have thrown fewer passes downfield. Dallas QBs have attempted passes of 10-plus air yards on 30.1 percent of throws this season, down from 35.2 percent in 2019; the team's air yards-per-attempt mark has also decreased by 1.4 air yards from last season to this season, the sixth-largest decline in the NFL.
And the decline isn't due to Prescott's absence, as the QB was averaging 7.9 air yards per attempt From Week 1 to Week 5 -- that's down significantly from his mark from 2019, when he averaged 9.3 air yards per attempt, the sixth-most in the NFL.
Now, why does this matter? According to Next Gen Stats, passes of 10-plus air yards average +0.40 EPA per play since 2019, compared to +0.10 EPA per play on passes of fewer than 10 air yards in that span.
Trusting the math: In-game decision-making
The in-game decision-making aspect has been the most obvious difference between McCarthy in Green Bay and McCarthy in Dallas. Through nine games for the Cowboys, McCarthy has attempted to go for it on 13.7 percent of fourth-down plays this season. This figure only trails the Jaguars at 15 percent. It's also 5.6 percentage points more than McCarthy's go-for-it rate over his final five seasons in Green Bay.
However, it's important to note that the NFL has become more aggressive overall in terms of going for it on fourth down. Consider that the Packers' 8.1 percent go-for-it rate on fourth downs from 2014 to '18 was the highest in the NFL in that span -- and that figure would be the 22nd-highest rate in the league today.
Still, in comparison to his predecessor in Dallas, Jason Garrett, McCarthy's been far more aggressive, going for it on a league-high 18 fourth downs, already blowing past the 2019 team's season total of seven. Although the results aren't there yet -- the Cowboys have only converted nine of those of 18 fourth-down attempts this season (50.0%, tied for 22nd highest in the NFL) -- the process is sound.
Notably, eight of their fourth-down attempts have come in that "sweet spot" on the field where coaches have difficulty deciding whether to go for it or punt: between their own 40-yard line and their opponent's 40-yard line. That's the most such attempts in the NFL -- and the Cowboys have converted five of those.
Additionally, McCarthy made noise for his mathematical prowess early on in the season, when he decided to try for a 2-point conversion against the Falcons at the beginning of what became a frenzied fourth-quarter comeback in Week 2. The attempt, coming after a touchdown pass with less than 5 minutes left that cut a 15-point deficit to 9, failed. But, as detailed by Pro Football Focus, the choice allowed the Cowboys to better shape their clock-management strategy for the final comeback push. When working to overcome an 8-point deficit, teams struggle to know whether it's better to slow down or hurry to secure an extra possession, but with a 9-point deficit, the Cowboys knew they had to move quickly to have a chance to win. Despite blowback from those who believed McCarthy should have kicked the PAT to prolong the game, the math and game theory behind the decision was sound.
Offensive creativity and stressing the defense
The most disappointing element of McCarthy's return has been the lack of creativity on offense, especially after he touted spending his time off studying every offensive snap of the top offensives in the league. When discussing his plans with Peter King long before he was hired by the Cowboys, McCarthy pointed to the Rams and 49ers as the NFL's smartest offensive teams due to their use of motion, snap-points and ability to frustrate defenses.
Yet, the Rams and 49ers, who are first and fourth in play-action percentage and seventh and first in shift and motion percentage since 2018 (per Next Gen Stats), do not resemble anything like this season's Cowboys offense.
Dallas has used play action on 20.2 percent of dropbacks this season, the 24th-highest rate in NFL. Their percentage was at 26.0 in Weeks 1-5, when Prescott was the team's starting QB, but that was still only good enough for the 13th-highest rate in the league. Perhaps this is in part due to game script, as the Cowboys have trailed early and often in their games this season. Teams do not like calling play-action when trailing in comparison to when they have a lead or are tied.
|Play-action by score differential||2019 Cowboys||2020 Cowboys||NFL average (since '19)|
|Leading or tied||35.4%||26.7%||27.7%|
But there may not be a discernible difference in play-action effectiveness based off score differential. In terms of yards per attempt, NFL teams average 8.6 yards per attempt on play-action passes when trailing, compared to 8.3 yards per attempt on play-action passes when leading or tied since 2019.
Additionally, McCarthy emphasized the use of shifts and motions to make the defense uncomfortable when talking to King.
"As a play-caller, you've got to stress the defense," McCarthy told King, "and one of the things watching all these teams has shown us is how good some teams are at challenging the eye discipline of the defense. Makes 'em think at the snap of the ball, which is huge. This bullet-motion sweep, this jet motion, at different tempos, different speeds. I just really like what it does to a defense. We call those things 'nuisances' for the defense."
Despite the use of motion being at the top of his list in terms of straining a defense in the offseason, McCarthy's Cowboys have decreased their shift and motion usage from 60.1 percent of offensive plays in 2019, the sixth-highest in the NFL, to 50.6 percent in 2020, the 13th highest. According to Next Gen Stats, the Cowboys' 9.5 percentage point decrease in shift and motion percentage is the fourth-largest such decrease in the NFL over the last two seasons. That said, 50.6 percent is still much higher than the Packers' 35.5 percent shift and motion rate in McCarthy's final season in Green Bay, which ranked 27th.
As forgettable as much of McCarthy's Dallas debut has been, there are some aspects to be encouraged by, like the increase in pass plays on early downs and additional aggressiveness on fourth downs. In short, his in-game decision-making has been solid. But overall, McCarthy must improve the team's creativity to match the claims he made in the offseason if the Cowboys are going to be successful in 2021 and beyond.