Support my attention-whoring ways by following us on twitter! https://twitter.com/StartKyleOrton

Get the SKOdcast imported directly into your brain! http://startkyleorton.podbean.com/feed/

Wednesday, January 2, 2019

Josh Rosen vs Josh Allen, or When "Stats vs Scouts" Ends in a Draw

Football in recent years has started, in fits and starts, to go through its own analytics revolution, similar to the one that overtook baseball in the early to mid 2000s. While the rise of sites like Football Outsiders, Pro Football Focus, and the introduction of the NFL's own house analytics page Next Gen stats, heralds a new and more informed way for fans to take in the game of football, it is clear that this revolution will not be as complete as baseball's. Football's requirement for teamwork more than individual excellence means that numbers alone will never tell anything close to the full stories. Quarterbacks are amplified or hindered by schemes, offensive lines, or butter fingered receivers. Cornerbacks benefit from elite pass rushers causing panicked and hurried throws or suffer from tepid pass rushes that allow QBs ample time to wait for the coverage to breakdown. While a number of new metrics try and break football down piece by piece to determine individual responsibility and performance on every play, it is clear there will always be a gray area when it comes to the story stats are telling about a football players performance, and how to project future performance from those stats. This is especially true when it comes to translating college performance to future NFL success.

No debate demonstrated the conflict between the rising tide of football analytics and the traditional methods of football scouting than the debate in the 2018 draft between quarterbacks Josh Rosen and Josh Allen. As previously noted on this blog there was zero, and I mean zero, statistical argument for drafting Josh Allen. Statistically he may have been the single worst QB taken in the first round in the last 20 years, at least since Kyle Boller. Scouts loved him, however, citing his underrated mobility, his zeus-hurling-thunderbolts level arm strength (he most definitely has what long-time readers of this blog will remember I once coined an "armcock"), his "leadership" and intangibles. Mel Kiper declared of course that "stats are for losers" and that Allen, most definitely, is a winner.

This avalanche of counterfactual, cliche-leaden, antiquated nonsense regarding what was clearly an unusually flawed prospect understandably appalled the more analytical minds of football media. In response they, too, found a cause to champion: Josh Rosen of UCLA. Rosen was himself a controversial prospect in the eyes of scouts, not because of his on field performance, but because off the field he was an outspoken liberal, a guy who appeared to value life outside of football, a unique personality willing to speak his mind and criticize coaches and teammates when he deemed it necessary. On the field Rosen was clearly Allen's statistical better in every category.

And so the stage was set: the grizzly old scouts, mouths full of chewing tobacco, car filled with old takeout containers from lifetimes spent traveling constantly to various backfields, evaluating players with gut instincts and finely honed senses locked into a culture war with basement-dwelling nerds who'd never picked up a ball thinking the entire game can be determined by spreadsheets. Whose Josh would win?



I am, of course, exaggerating somewhat but it was clear which player the two camps backed and why. So one year down, how has the argument played out? Surely one side has proven the validity of their argument once and for all?

Well...uhhh...

That's ANYA, by the way, Pro Football Reference's excellent metric which takes yards per attempt and adds bonus points for touchdowns and takes away points for interceptions and sacks. The stat isn't perfect (no stat is), but it is the passing # that correlates best to winning.

So, okay. Both Rosen and Allen sucked just going by the raw numbers. But they're rookies! Why Sam Darnold's just above them if you pull up the query yourself. Defenders of both will make the same arguments about poor supporting casts and/or schemes, they'll argue neither is a finished product. All of this is true, and makes sense, of course. Except....

Well, most first round QBs are drafted by terrible franchises. That's why they need QBs. Most get stuck behind subpar lines, throwing to subpar receivers, often with inept coaches who will soon be fired for their failure to develop. So it's fair to ask why these two suck so much worse than their peers.

Using ANYA+, PFR's stat that adjusts ANYA for era (the league has become much more pass friendly, so averaging just 5 ANYA in 2018 is a far more egregious sin than doing so in 1999), Rosen and Allen rank 64th an 46th overall respectively out of a sample of 64 first year QBs with at least 200 pass attempts in the last 20 years. If you're a Bills fan and you're encouraged that Allen at least is 18 spots higher on the list than Rosen, consider that Allen's particular shortcoming is his accuracy, and in terms of completion % + (completion % relative to league average adjusted for era just like ANYA+), the only QB who was less accurate as a rookie than him was Ryan Leaf himself.

Are there other examples of QBs performing as poorly in terms of ANYA+ as Allen and Rosen and turning it around to have successful careers? There are just three, actually: Matthew Stafford, Donovan McNabb, and, of course, the example most often cited by partisans for both Allen and Rosen: Jared Goff. I'm not convinced, however, that the quick turnarounds those three QBs pulled off will be possible for Allen and Rosen, however.

In Allen's case, the reasons are quite simple. Allen isn't an accurate passer. He hasn't completed 60% of his passes at any level, even going back to high school. There's simply not a single example in modern NFL history of such an inaccurate passer suddenly developing that skill at the pro level. He will continue to show flashes of greatness, like his recent game against the Dolphins, but it's beyond unlikely that he will ever manage the consistency required to justify his draft slot . If Bills fans don't want to believe me, remember that even Kyle Boller had 12 different games in his career with a passer rating above 97, and 5 different games where he threw for 3+ TDs. Each was touted as the game where he "finally put it all together" only to be followed soon after by a regression to the mean.

The Goff analogy also fails for Allen because Goff didn't just naturally progress from the disaster of his rookie year to the top 10 passer he is now. The Rams overhauled their entire organization and hired one of the brightest offensive minds in football to fix Goff's deficiencies. The Bills, on the other hand, seem quite pleased with the job HC Sean McDermott and OC Brian Daboll have done with Allen, and are likely to hold their jobs next year. No quick schematic fix is coming there. The best case scenario for Allen would be the Donovan McNabb scenario, where he, like McNabb, improves his accuracy somewhat while remaining a below average completion % player nonetheless but makes up for it with big plays downfield and an ability to avoid interceptions. Unfortunately, nothing in Allen's college profile suggests he can do that, as he was intercepted on 3.2% of his throws and completed just 56% of his passes for only 7.7 adjusted yards per attempt, all pedestrian numbers especially when compared to McNabb, who completed 58% of his passes in college (but 62.5% in his final year vs a final year tally of 56.2 for Allen) while routinely hitting big plays for an adjusted yards per attempt figure of 9.3 and an interception % of just 2.8, and McNabb did this over 20 years ago in a much less pass-happy college environment. There's simply no comparison to be made, frankly.

But what of Rosen? Surely he's a good fit for the Goff comparison. He's a smart, polished passer, no? Much more so than Allen. It's just an excuse when Bills fans blame Allen's supporting cast, but it's clearly true for Rosen, right?

Well, the underlying numbers don't like Rosen much better, for one. Rosen's actual completion % of 55.4 is 4.2% worse than his expected completion % of 59.4 (expected completion % takes into account drops, depth of target, how open the receiver was, etc to determine true accuracy), which was 5th worst in the NFL. Allen was 3rd worst in the NFL with a 6.8% gap between his xComp% of 59.6 and his actual completion % of 52.8. ESPN's QBR (which takes into account drops, throwaways, depth of target, distance to the marker, etc) actually deemed Rosen an even worse passer than Allen, with Allen their 31st ranked passer and Rosen their 33rd.

Simply put, the accurate, polished passer that Rosen was praised for being in college hasn't shown up in the NFL so far. His supporting cast is indeed terrible, and I predict he has improvement in his future, but I would caution anyone going immediately to the Goff comparison here for several reasons as well.

For one, the Cardinals offense is an utter mess bereft of playmakers outside of David Johnson, who may or may not be the same explosive player he was pre-ACL tear, and it's unclear who, if any, of Arizona's starters on the OL might be expected to be part of a future, more competent line. The 2016 Rams actually had the core of their current offensive line already on the roster, and David Gurley available to be a workhorse back. While the Rams have added to the OL and added pass catchers, there appears to have been less work to do for the Rams to put a competent offensive unit around Goff than the Cardinals have for Rosen.

For another, there's only one Sean McVay. Sure, there's a number of bright offensive minds out there, and McVay hardly invented offense. One need only look at Doug Pedersen in Philadelphia and Matt Nagy in Chicago and Anthony Lynn in San Diego to see there are other guys out there capable of transforming a franchise through a new offensive philosophy, but guys like McVay who can take a completely moribund offense and transform it into a top 10 attack are so incredibly rare as to be an almost certain aberration. Lots of teams are going to try to find the next McVay/Nagy and plenty are going to whiff and land the next Marc Trestman or Ben McAdoo. If you're making your entire QB reclamation project dependent on landing a literal prodigy, maybe you've erred already.

Finally, I'm not convinced Josh Rosen is that good of a prospect to begin with, and I think he benefited from the analytics crowd needing their own champion to counter Allen that they overlooked some red flags themselves. For one, Rosen wasn't super-productive in college. His career completion % in college ranked just 41st out of 60 first round QBs taken since 1998. Is that better than Allen's abysmal 55th? Certainly, but its hardly in "can't miss" territory, and in fact even the handful of passers who were less accurate than Rosen in college who became successful NFL QBs regardless mostly achieved their success as vertical, big play passers who were nonetheless less accurate/efficient than the league average. Rosen is unlikely to ever be a major downfield passer in the NFL, however, as he again ranked just 41st of the 60 first round QBs in terms of adjusted yards per attempt. Scouting reports frequently made note of Rosen's arm strength being merely OK for a first round QB, and that more often than not he needed to have perfect protection and a chance to step into his throws to really push the ball downfield. Rosen's strength was supposed to be timing, anticipation, grace under pressure, and accuracy, which makes the extent of his struggles in less than ideal circumstances a bit concerning.

Do I think Rosen is necessarily as doomed to fail as Josh Allen? Not necessarily. I am sure with a better scheme and supporting cast he will at least improve as a passer, while I expect Josh Allen to remain a roller coaster ride until the end. That said, I think a lot needs to go right in order for Rosen to ever reach his ceiling, which I think will ultimately be more Joe Flacco than Peyton Manning. Considering we're talking about a dysfunctional organization that just fired a coach after a single year, I have my doubts Rosen will ever get there.

In the end, the Rosen-Allen debate may end up being football's version of Juan Pierre vs Adam Dunn. Pierre, the hard scrabbled, base-stealing, high contact/average, max effort center fielder was the darling of the baseball old guard, while Dunn, a three true outcomes hitter who was laughably terrible in the field, was beloved by early sabermetrics advocates who scoffed at the idea that baserunning or defense was so valuable that Dunn's deficiencies in those areas outweighed the value provided by his ludicrous power and on base skills. In the end, however, newer metrics like Wins Above Replacement and advanced defense and baserunning metrics showed that Dunn and Pierre, though vastly different in approach, were largely equally valuable as more or less average baseball players. In Josh Allen and Josh Rosen we have one athletic freak of nature whose highs and lows will likely never flatten out into a consistently above average player, while Rosen's merely adequate physical skills and better, but ultimately unimpressive college statistics, might result in an equally mediocre QB nonetheless.

Perhaps, as the best baseball front offices learned after a decade of asserting analytics preeminence over old-school scouting and conventional wisdom, football front offices (and those who write aboutthem) will learn to look for where scouting and statistics reach consensus rather than conflict. 

No comments: