Check out Quarterback Scarcity is Over, Part 1.
Josh Allen should not be this good. Allen went to a small school (Wyoming) where he completed under 60% of his passes for his career. In 2017, his final season, he ranked 103rd out of 139 qualifying FBS quarterbacks in raw completion percentage. It’s true that at small schools like Wyoming, Allen wasn’t surrounded by the greatest supporting talent, but it’s also true that he was facing inferior defenses and his inability to light up lesser competition was concerning.
Allen, who was drafted 7th overall as the third quarterback taken in the 2018 draft, is in many ways the prototypical bust. He’s a physical specimen, extremely fast, with the strongest arm the NFL has seen in years. Quarterbacks like Allen are often drafted too high based on the idea that NFL coaching can harness that raw physical ability and fix whatever mechanical and accuracy issues may exist.
In reality, this almost never happens. The best statistical indicator of future NFL success, Completion Percentage Over Expected (CPOE), indicates that one’s accuracy is more of a fixed trait than something you can improve. NFL Combine superstars who were inaccurate throwers in college litter the list of highly drafted busts as sure as Kyle Boller can hit an upright from his knees from 60 yards out, and most stat crunchers assumed Allen would be joining the list.
Allen did nothing to prove analysts wrong in his first two NFL seasons, finishing 33rd in DVOA in 2018 as a rookie (-35.9%), and improving slightly to 28th in 2019 (-11.8%). For context, Mitch Trubisky finished ahead of Allen in 2019 with a -11.0% DVOA. For quarterbacks with at least 200 attempts in 2019, Allen finished 32nd in completion percentage with 58.79, barely improving upon his college mark of 56.2%. He finished 23rd in ANY/A just behind Ryan Fitzpatrick. If there was a Josh Allen boom coming, there was scant evidence for it on the field.
Then a funny thing happened. Josh Allen showed up in 2020 an entirely different player. It’s not just that his statistics drastically improved (though they did); his throws looked much crisper, especially in the intermediate and deep spaces. Allen, who previously had trouble throwing consistently catchable balls, was now not only hitting his receivers, but locating throws on specific sides of his target. And while there are some external reasons for his improvement, most prominently the addition of Stefon Diggs, it quickly became clear that Allen was a changed player.
Allen finished 2020 as a top MVP candidate, losing out to Aaron Rodgers, but had he won it would have been tough to complain. Allen’s completion percentage soared to an unheard of (for him) 69%, 4th overall and just a point behind Rodgers. While Allen finished 6th in ANY/A, no quarterback ahead of him could rival him as a rushing threat, and his total value was almost certainly not significantly different than Rodgers or Mahomes. Allen rushed for 421 yards and 8 touchdowns, his 3rd consecutive season with at least 8 rushing touchdowns. Allen finished 3rd in passing DVOA and while his rushing DVOA was only 18th, his DYAR was 4th. No other quarterback was as big a part of both the running and passing game than Allen, and I’m still not sure that any quarterback has ever taken quite as big a leap from his second year to his third.
Whenever a quarterback, or any player takes a leap like this, it’s natural to ask if it’s sustainable. After all, Carson Wentz jumped from a DVOA of -12% in 2016 to a 6th ranked 23.8% in 2017, but has since crashed completely. Given how good the Bills’ coaching staff and offensive play-calling has been, it’s fair to ask if Allen would succeed outside of the current environment.
I’m confident that this improvement is for real, because we have total agreement between scouts, who see Allen’s improved technique, and statistics, which have no obvious regression points. Frank Reich’s scheme and unsustainable results on 3rd down and close to the goal line in Philadelphia are the primary reasons Wentz was successful, which is why the Eagles were able to bring a championship home with backup Nick Foles when Wentz went down. But with Allen, the change is clear.
So how exactly did this happen?
To answer that, let’s take a quick detour to the world of baseball, and Driveline. Driveline is a player development system created by Kyle Boddy, a data scientist-turned-pitching guru who once worked for the online gambling site PokerStars to identify flaws in their security. Boddy’s work in creating Driveline led him to a big league job, as he currently works as a trainer and consultant for the Cincinnati Reds.
Boddy’s story is well documented elsewhere, including in the recent book The MVP Machine by Travis Sawchik and Ben Lindbergh. The MVP machine focuses primarily on the odious Houston Astros, and the perhaps more odious reigning Cy Young award winner Trevor Bauer, but no one should let the Astros “banging scandal” or Bauer’s propensity to make enemies detract from the fact that Driveline gets results. The MVP Machine is, more than anything, a treatise on how technology can revolutionize player development. After all, say what you will about Bauer, he did win his Cy Young.
Driveline uses a mix of high-speed cameras and a unique training methodology involving weighted balls, which reliably adds 1-2 MPH to a given fastball or RPMs to a given breaking ball. Weighted ball regimes are also useful in that they appear to be safer than throwing regular baseballs. Boddy actually had this insight because of football, because quarterbacks almost never experience the arm injuries that pitchers due despite throwing a much heavier ball. The weighted balls regimen also teaches pitchers better command, as they learn to hit their spots despite the variable of the changing ball.
But the hallmark of Driveline is focused practice geared toward actionable improvements. The real innovation Boddy made wasn’t specifically about pitching, but in having specific measurable goals and evidence to support the efficacy of his methods. At Driveline, they do not use old-timey drills with a long history and tradition. Cameras, radar guns, and biomechanical experts are used to identify specific weaknesses in a pitcher’s delivery, and the corresponding drills assigned to fix those weaknesses are vetted, measured, and proven effective. Every drill has a purpose, and for that reason, every drill works if the player puts in the effort.
Aside from weighted balls, the other famous tool of Driveline is the high speed Edgertronic Camera, capable of shooting 500 high definition frames per second, or, if you want to sacrifice some fidelity, up to 17000 frames per second. The Edgertronic allows for an exact breakdown of spin rate on the ball, as well as any minor mechanical imperfection of the pitcher. Once you can identify issues with precision, it’s just up to a coach to figure out how to fix it. Some issues, of course, can’t be fixed. Innate talent still counts for something. But, given enough information, time, and most importantly, drive from the athlete involved, most issues can.
Back to football. The fact that players can be developed should be no surprise to anyone. After all, Aaron Rodgers was quite famously a mechanical mess coming out of college, and if there is a poster child for the quarterback development track before Allen, surely it’s Rodgers. The difference is in what coaches were able to fix. NFL coaches have always excelled in fixing mechanical issues and getting a few extra MPH out of a delivery. Rodgers may have been a mess, but unlike Allen, he was a naturally accurate passer in college:
Allen, on the other hand, struggled with his accuracy, a skill that generally had been beyond development. As it turns out, with enough information, not much is beyond development.
Jordan Palmer, younger brother of Carson, attempted 18 NFL passes over four seasons, completing 11 for no touchdowns and two interceptions. While Jordan managed to make a few million dollars over his tenure, he was a career backup and never threatened to make the leap as a starter. That said, time spent in NFL locker rooms working under a variety of coaches with a variety of quarterbacks of varying quality is itself valuable. The old saying goes that those who cannot do, teach. It’s a worthless cliché, as teaching is its own hard-to-come-by skill, and Palmer likely wouldn’t be where he is as one of the NFL’s premiere quarterback development specialists without his brief stint in the league.
We also shouldn’t be so quick to discount Palmer’s results as a player entirely either. Palmer was a four year starter for UTEP in college and over his tenure he improved methodically from season to season until, as a senior, he was one of the most efficient passers in football.
If there was a player out there who might understand how to improve accuracy, which drills were effective and ineffective, and how to relate to the unique egos that occupy the QB position (both from his time in the game, and dealing with his own brother) it would be Jordan Palmer.
Palmer made offseason headlines when reporters began trying to explain why Allen was able to improve. Most roads led to Palmer, and specifically to his beach workouts. Nora Princiotti wrote an excellent piece for The Ringer detailing why these workouts were important:
“You push away, and your foot slides a little bit, and your left foot is on a hill that you didn’t really see, so every single time your feet are jacked up,” Palmer said. “What happens is you have to fix things from your hips up, in your upper body, and you have to get really consistent there. He plays with a really consistent base now.”
Just as Driveline uses weighted balls to create velocity and command under changing circumstances, the deep, dry sand at the beach creates uncertainty from throw to throw. In order to remain accurate in such an environment, upper body mechanics need to be detached from the lower body. One of the keys to good mechanics, and for making the most of a dynamic scrambler like Allen, is to create repeatable upper body mechanics even when the lower body is scrambling, being hit, or otherwise compromised. The hallmark of a modern quarterback is accuracy out of the pocket (or under pressure) as much as in the pocket, but developing that skill isn’t as simple as simulated pressure or repeated practice on rollouts. What the beach gives you is real unpredictability, and the only way to improve without solid footing is to create new habits for your upper body.
Many of Palmer’s other drills also involve unpredictability, and his espoused philosophy is right in line with the Driveline ethos of focused practice creating demonstrable results.
Palmer has worked with many clients in preparing for the draft including Allen, Patrick Mahomes, and Deshaun Watson, but they’re not all success stories. Allen has continued to work with Palmer as a pro, but so has Sam Darnold. Darnold hasn’t shown the improvement of Allen, though it will be interesting to see what happens should he find himself in a better environment. The point of this isn’t that Palmer is some quarterback savant; rather, the point is that his approach to coaching prospects is progressive and modern. He determines what he has in each prospect, and designs a regimen for a specific problem.
If you dig into Palmer’s philosophy a little deeper (for research purposes, I signed up to receive QBSummit emails from his company), there is some consultant-speak rah-rah babble that comes along with Palmer, but it’s undeniable that he’s an effective cheerleader and his focus on diligent practice habits and behavioral red flags is a big part of being an effective coach. It’s all well and good to be able to create effective practice routines and scientifically analyze the results, but if the prospect in question isn’t willing to put in the work, it doesn’t matter.
Bringing this back to Aaron Rodgers for a second, he is, prior to Allen, the prototype for a developmental prospect, and in retrospect it’s not shocking, because Aaron Rodgers is clearly willing to put in the kind of work necessary to become the greatest. Rodgers has that Michael Jordan-style edge that often rubs people the wrong way, and he gets on his teammates for poor execution though he is also quick to praise those who turn a corner, as with Marquez Valdes-Scantling this year. If you tell a player like Rodgers how to get better, he will. Allen is by most accounts the same way.
When scouting quarterbacks then, the best indicator of their upside is almost certainly in “makeup.” This quality is often impossible for outsiders to judge, but we’re also not totally in the dark. A player like Johnny Manziel seemingly did not have the work ethic necessary to turn himself into a high-level player. We should probably pay more attention to red flags than we do.
Tape is Dead, Long Live Video
The NFL is a pioneer in the use of video, and any scout, coach, or even fanboy wannabe looking to prove his mettle need only cite the hours of tape they grind and the use of “All-22” to put their credibility out there. NFL coaches have used tape sessions for almost as long as they’ve had film, and generally speaking, they’re quite good at it. They have also enjoyed a virtual monopoly on tape availability until recently, but that’s changing fast. The NFL is letting more out the door than ever before including their precious All-22 via the NFL Gamepass, and through the gradual publication of tape-based statistics via Next Gen Stats. We also have outside organizations like Pro Football Focus and Sports Info Solutions selling in-depth tape quantification to any fan or analyst who wants it.
Back to baseball for a second. One of my favorite recent developments in baseball is the “juiced ball” going from a conspiracy, only acknowledged as a possibility by kooks, to an area of sabermetrics that we now understand with incredible precision. Much of this is due to the efforts of Rob Arthur, who does his own yearly quantification of the juiced level of the ball, and collects data from the best and brightest studying the issue. Arthur writes for Baseball Prospectus and FiveThirtyEight, and the creativity and drive Arthur shows to understand the baseball using frame by frame tape breakdown, X-rays, and the physics of drag as it relates to seam height is incredible. Give a quick read to this 2018 piece by Rob and collaborator Tim Dix. Keep Dix in mind as he’ll be showing up again momentarily.
The sophistication of outside analysts in baseball is unparalleled across sports, and their research here has put Major League Baseball into an uncomfortable position of acknowledging that the makeup of the ball matters very much indeed, and making public statements (even if unintentionally) around their “ball strategy” for the coming year. It’s frankly insane, and it was all driven by sophisticated video tracking and applied physics by people outside of the game.
Back to football, just as Driveline uses the Edgertronic camera to inform efficient practice techniques, amateur analysts are getting better and better at using a combination of video and various player tracking and analysis systems to quantify the game more than ever before. You need only look at the most recent Big Data Bowl, which resulted in several new and better ways to quantify wide receiver efficiency, or with the NFL Draft upcoming, or this thread on the actual impact of a receiver’s raw speed on their ability to play in the NFL:
Let’s see if I can not mess this thread up...— cynthia frelund (@cfrelund) March 7, 2021
My findings from computer vision derived on-field WR speeds.
First, what I did: measured 8 seasons of on-field (game) college WR speeds (with pads), sorted by route.
Mapped them to 40 times from #NFLCombine
I mention this last one because it’s becoming more and more clear that as we gain sophistication measuring athleticism on the field through frame-by-frame tape analysis, data derived from the NFL Combine and college pro days become less and less valuable. Quarterbacking is no exception.
On Jordan Love’s projection
Let’s talk about Jordan Love. Most highly drafted quarterbacks end up starting their rookie year, at least if they are any good. Love and Rodgers are exceptions as quarterbacks who were slated to develop on the bench. Given what we know of Josh Allen, who became a project quarterback based on his incredible athleticism, I think it makes sense to take a look at whether Love shares any of what makes Allen special.
Are you interested in Jordan Love’s arm strength compared to Josh Allen, or to his draft class contemporaries? A radar gun will give you an extremely incomplete picture, as quarterbacks rarely throw under circumstances similar to a pitcher, but frame-by-frame tape breakdown is providing us more and more insight into things like arm strength and overall delivery time. Remember Tim Dix from the baseball article a few paragraphs ago? Video gurus like Dix are producing frame-by-frame breakdowns of quarterbacks to produce this data, and to quantify where any struggles may lie:
“As it says in the title, I’ve counted the frames and “done the math” on more than 400 throws from the players who I deem to be the top-four quarterback prospects in the 2020 NFL Draft: LSU’s Joe Burrow, Alabama’s Tua Tagovailoa, Utah State’s Jordan Love, and Oregon’s Justin Herbert. Now, if you just want to know how much faster Tua’s release is than Joe Burrow’s, or which QB has the strongest arm, skip ahead to the “RESULTS” section. And if you really don’t feel like reading, there’s a TL;DR at the bottom of the page. But if you’re even slightly curious how these results were calculated, I’m going to briefly explain how all of this works before diving into the data.”
Dix applied a process common in baseball to football, measuring frame rates as a proxy for speed, allowing him to also account for the time of the throwing motion as well as “trigger time,” or the time between locking onto a target and beginning the throwing motion. By doing so, Dix achieved a novel and useful breakdown of where functional quarterback velocity comes from. As it so happens, Jordan Love has the strongest arm in the class per Dix’s tape analysis, and by a healthy margin. This is not a controversial opinion, and Love is commonly known to possess elite arm strength. Where he struggles is in “trigger time,” and here we see how the most useful comp for Love may in fact be Josh Allen.
In college, Love and Allen both took quite a while to lock onto their targets, relying on that arm talent to substitute for anticipation. But relying on overthrowing to compensate for anticipation will cause your accuracy to suffer in the aggregate, make things harder on your receivers, and lead to seasons like Love’s final year in which he was an interception machine.
Love also possesses a similar physical skill set to Allen. Just take a look at their RAS cards, paying specific attention to their weakness in short shuttle. Here’s Love:
I’ve seen Love compared to Patrick Mahomes far too often, and that’s not correct or fair. The Packers obviously had no interest in playing Love immediately with Rodgers still on the team, and it’s clear they drafted him specifically to develop. He possesses an absolute rifle, he’s athletic in the same way as Allen, and he shares many of Allen’s college weaknesses. The comp here for Love, at least as a ceiling, is Allen. At this point, for Love, it’s a question of makeup.
Dix is obviously limited by what is available to consumers, but NFL teams have no such limits. If analysis this sophisticated is coming from fans with a passion, Driveline-style Edgertronic analysis combined with the information available behind the scenes at Next Gen are giving sophisticated NFL teams more video, and more data based on that video, than ever before. Modern quarterback development combined with modern NFL offense has led us to some of the most entertaining football ever played, and really, we’re just getting started.
Stay tuned for part 3 in the coming days.