Stat Analysis
Advanced analytics on player and team performance

QB Functional Mobility Model 2020

Utah State Aggies QB Jordan Love
Photo: USA Today Sports Images

Guest column by Jeremy Rosen and Alexandre Olbrecht

2019 was the year of the mobile quarterback. In April, the Arizona Cardinals drafted Kyler Murray (who received the highest projection in our model last year) with the first overall pick. Murray then won Offensive Rookie of the Year and became the second first-year player in NFL history with at least 3,500 passing yards and 500 rushing yards (Cam Newton was the first). Moreover, his 7.1 rushing yards per attempt in college were the highest of any quarterback drafted in the first three rounds dating back to at least 1985. Not far behind him, with 6.3 rushing yards per attempt in college, was Lamar Jackson. All Jackson did in 2019 was rush for an NFL quarterback-record 1,206 yards, post a passer rating of 113.3, and win league MVP.

However, as shown in Table 1, our functional mobility model's projected Adjusted Net Yards per Attempt (ANY/A) for Jackson was only 5.21, squarely in the middle of our projection range. It was relatively low because, as mobile as Jackson was in college, he lacked the functional part. A functionally mobile quarterback runs effectively but rarely, whereas Jackson was a dual threat with more rushing attempts than completions. So how did he become that good? On one hand, his incredible arm and leg talent make him an outlier. But on the other hand, he is far from the only recently drafted quarterback to succeed via his mobility.

Table 1: 2015-2019 Projections
Quarterback Projected ANY/A NFL ANY/A
Marcus Mariota 6.27 6.12
Kyler Murray 6.16 5.55
Drew Lock 5.78 6.20
Mitchell Trubisky 5.78 5.56
Sam Darnold 5.75 5.37
Davis Webb 5.64 3.40
Dwayne Haskins 5.58 4.25
Carson Wentz 5.49 6.30
Jameis Winston 5.49 6.30
Deshaun Watson 5.44 6.83
Patrick Mahomes 5.37 8.57
Will Grier 5.36 3.40
Baker Mayfield 5.28 5.99
Lamar Jackson 5.21 7.52
Paxton Lynch 5.21 4.10
Jared Goff 5.16 6.73
Daniel Jones 5.15 5.38
Josh Rosen 5.14 3.40
Garrett Grayson 5.13 3.40
DeShone Kizer 5.09 3.49
Mason Rudolph 5.05 5.02
Jacoby Brissett 5.02 5.63
Josh Allen 4.85 5.16
C.J. Beathard 4.81 4.74
Cody Kessler 4.52 4.56
Christian Hackenberg 4.46 3.40
Sean Mannion 4.10 3.40
Correlation w/ NFL ANY/A: 0.373 (p = 0.028)

Ryan Finley and Jarrett Stidham are not in Table 1 since they were not drafted in the first three rounds. In addition, as in the previous two years, we set a minimum ANY/A equal to Jimmy Clausen's 3.40.

Revisiting the Classes of 2018 and 2019

When we built our model with data of quarterbacks drafted from 2000 to 2014, the numbers were clear. The best predictor of NFL success was functional mobility, which we measure with college rushing yards per attempt (ideally high) and the natural logarithm (ln) of college run-completion ratio (ideally low). We are happy to report that, per Table 1, functional mobility remains significantly positively correlated with NFL success. However, the mobility part, not the functional part, is responsible for most of functional mobility's predictive power, especially recently. To demonstrate, in Table 2, we list the Classes of 2018 and 2019 from highest to lowest college rushing yards per attempt, along with each quarterback's NFL ANY/A. With a correlation coefficient of 0.711, mobility in college has been incredibly predictive of NFL success.

As such, the question now is whether functional mobility should give way to pure mobility. We present four arguments in favor of pure mobility and four against it:

In favor of:

(1) Table 2 shows a clear-cut trend: mobile quarterbacks are succeeding in the NFL.

(2) Unlike in the past, some NFL playcallers, such as Greg Roman, are building their offenses around such quarterbacks.

(3) By sliding and avoiding hits, quarterbacks such as Russell Wilson can scramble without getting injured. They also benefit from stricter enforcement of rules designed to protect quarterbacks.

(4) Now that running is less stigmatized, top prospects may be willing to run more often in college. Therefore, a high ln(run-completion ratio) may no longer be a red flag.

Against:

(1) The past two years are a small sample size. In hindsight, pure mobility was also a better predictor than functional mobility from 2015 to 2019, but even that sample is small.

(2) Given the success of mobile quarterbacks in 2019, NFL defenses will focus on scheming ways to stop them.

(3) Even if such quarterbacks avoid serious injuries, they will slow down later in their careers.

(4) Pocket passing is a skill that takes time to develop. Dual-threat players, such as Jackson and Josh Allen, are succeeding early, but more traditional passers, such as Baker Mayfield and Sam Darnold, have plenty of room for growth.

Table 2: 2018-2019 Mobility
Quarterback College RY/A NFL ANY/A
Kyler Murray 7.1 5.55
Lamar Jackson 6.3 7.52
Daniel Jones 3.3 5.38
Josh Allen 3.2 5.16
Baker Mayfield 2.7 5.99
Sam Darnold 2.4 5.37
Drew Lock 2.2 6.20
Dwayne Haskins 1.9 4.25
Will Grier 1.0 3.40
Mason Rudolph 0.1 5.02
Josh Rosen -1.4 3.40
Correlation w/ NFL ANY/A: 0.711 (p = 0.007)

Projections for the Class of 2020

Given that functional mobility remains predictive of NFL success, and the case for pure mobility to replace it is unsettled, we have retained the same model for 2020. Having said that, we have updated our dataset with the most recent data. In addition, 95% PI stands for 95% Prediction Interval; each quarterback's career Adjusted Net Yards per Attempt (ANY/A) has a 95% chance of ending up in his 95% PI. Because there is a lot of uncertainty with quarterback prospects, these intervals are necessarily wide.

Table 3: 2020 Projections
Quarterback School Projected ANY/A 95% PI Low 95% PI High
Tua Tagovailoa Alabama 6.11 3.98 8.24
Jordan Love Utah State 5.74 3.64 7.85
Justin Herbert Oregon 5.63 3.53 7.73
Joe Burrow LSU 5.50 3.41 7.60
Jalen Hurts Oklahoma 5.34 3.22 7.46
Jake Fromm Georgia 5.19 3.10 7.29
Jacob Eason Washington 4.80 2.70 6.90

We start with Heisman Trophy winner and presumptive first overall pick Joe Burrow. He had one of the greatest seasons of all time by a college quarterback and would have a projected ANY/A of 5.87 if 2019 were his only season. However, he also started for LSU in 2018, when he was not nearly as successful. Our model is not weighted in favor of more recent seasons; if it were, his projection would be higher.

Meanwhile, our highest projection goes to Tua Tagovailoa, who is the quintessential functionally mobile quarterback. Unfortunately, he fractured his hip last November, and our model does not factor in the risk of that injury derailing his NFL career.

Rounding out the top-tier prospects, per Scouts Inc., are Jordan Love and Justin Herbert. These two quarterbacks have similar functional mobility profiles. Neither is as mobile as Burrow or Tagovailoa, but both have a lower ln(run-pass ratio) than Burrow.

Finally, the second tier of prospects consists of Jake Fromm, Jacob Eason, and Jalen Hurts. Unlike the top-tier prospects, Fromm and Eason are traditional pocket passers. While neither is particularly mobile, Fromm is still significantly more accurate and mobile than Eason, who has a completion percentage below 60% and -1.5 rushing yards per attempt. Since 2000, nine other quarterbacks drafted in the first three rounds have had that few or fewer rushing yards per attempt, and the only one with any NFL success is Nick Foles. Lastly, Hurts is 2020's true dual-threat quarterback; he has almost as many rushing attempts as completions, but he also has 5.3 rushing yards per attempt.

The future of functional mobility

Lamar Jackson's MVP award was a major win for mobile quarterbacks, and the early-career struggles of Will Grier, Mason Rudolph, and Josh Rosen cast doubt on the prospects of immobile pocket passers. However, given the sky-high potential of Burrow and Tagovailoa, the middle ground of the functionally mobile quarterback is likely here to stay. Having said that, we are most intrigued by what will happen with Hurts, in both the draft and his career. Unlike Jackson and Kyler Murray, many scouts do not view Hurts as a particularly talented passer. Therefore, if he becomes a successful NFL starter, that may indicate a complete paradigm shift in the way the league views quarterbacks.

Jeremy Rosen conducted the majority of this research, which is now published in Contemporary Economic Policy, at Georgetown University. Alexandre Olbrecht is an associate professor of economics at Ramapo College of New Jersey and the Executive Director of the Eastern Economic Association. The views in this column are entirely our own and do not represent those of Contemporary Economic Policy, Georgetown University, Ramapo College, the State of New Jersey, or the Eastern Economic Association. Credit to Doug Farrar for his SI.com article on functionally mobile quarterbacks.

Comments

5 comments, Last at 04 Apr 2020, 11:30am

1 A functionally mobile…

A functionally mobile quarterback runs effectively but rarely, whereas Jackson was a dual threat with more rushing attempts than completions. So how did he become that good?

This seems to be begging the question, where the question assumes the answer.

Looking at the list of over-achievers and under-achievers, I can't help but think I see a coaching effect (or overall roster talent). Nearly all the big losers were in dumpster fires, whereas the big winners were paired with good teams/good coaches. Is Pat Mahomes the big winner here if he was playing on some Fisherian 7-9 bullshit?

2 Agree with this assessment. …

Agree with this assessment.  If the Jets were smart, and sat Darnold like they did with their last two franchise quarterbacks, (O'Brien and Pennington), he's in a much better place now.  Honestly, Mayfield would have done better if he had competent coaching, but he's actually done alright all things considering (negative DVOA last year but barely positive DYAR, and solid numbers his rookie year).  Jackson and Mahommes have had the best luck with coaching, and we'll see if Murray is in a good situation or if last year was a fluke.

3 As a once practicing…

As a once practicing econometrician(and I confess I have not yet read the paper), there is endogeneity baked into these numbers. Does the scheme skew the qb's statistics vs the opposite. Depending on which direction causality runs, you are going to get very biased estimates.

5 "Dual-threat players, such…

"Dual-threat players, such as Jackson and Josh Allen, are succeeding early," by what measure is Josh Allen (-21 DYAR) succeeding?