clock menu more-arrow no yes mobile

Filed under:

Scrambling for a Quarterback: Parcells’ Checklist

Every year, teams desperately try to figure out how to draft a good quarterback. Bill Parcells had a list of traits he used to help him. Does that list actually work?

NFC Divisional Playoffs - Chicago Bears v New York Giants Photo by Focus on Sport/Getty Images

Bill Parcells supposedly had a checklist for what to look for in a college quarterback in order to be draft-worthy. He wanted a player who was a three-year starter, a graduating senior, and a “winner” with at least 23 wins (out of at least 30 starts). The player needed a 2:1 touchdown to interception ratio and at least a 60% completion rating.

I wanted to test Parcells’ checklist, and I wanted to see if I could do any better. However, in the interest of full disclosure, I came into this project highly skeptical. For all that I am a “stats” guy, and for all that I love analytics, applying such an approach to the draft is full of problems. Statistics only tell part of the story, and that’s more true in college football (with its different talent levels, conferences, coaching abilities, and team compositions) than can be accounted for by any casual fan—or even by most experts.

Therefore, when it comes to the draft, I am a “watch the tape” guy and not a “scout the boxscore” guy. However, I tried to keep an open mind.

I pulled ten years of drafted quarterbacks from 2009-2018. This meant that even the players drafted in 2018 had three full seasons to be evaluated. Any quarterback who was drafted went into the main pool, and I pulled data available from major reference databases (mostly Pro Football Reference and Sports Reference, but I filled in the blanks with some school databases, too, when they were available). This did mean that I had to leave out Parcells’ insistence that a quarterback graduate—there is no single database for that, and even public profiles for these players seldom indicate graduation status.

As a college educator, I also have to point out that Parcells’ insistence on college graduates is interesting. Only about half of NFL athletes have graduated from college, and whether or not Parcells meant to create this division, he placed a tremendous amount of importance on a factor that is more a result of socio-economic privilege than any trait inherent to the player. Besides, of the 21 quarterbacks in this group to have earned a Pro Bowl, only 10 of them even played their senior year, so this requirement is sort of a non-starter. However, with that disclaimer out of the way, I measured Parcells’ remaining six traits against starts, pro bowls, snap counts, median indexed passer rating, and career adjusted net yards per attempt.

Overview

Ten years of draft picks gave me 118 quarterbacks, and even trimming out the quarterbacks who had no college stats available left me with 112 players. That’s a pretty robust data set for the circumstances. Just measuring how many of Parcells’ desired traits a player has and comparing it to his career adjusted net yards per attempt results in a correlation of 0.20. That’s really low. While a 0.80 is a strong correlation and a 0.60 is still worth talking about, a 0.20 is basically a “are you sure you entered the right numbers or did your cat jump on your keyboard?” level of disconnectedness.

32 of these players passed all six remaining elements of the Parcells test, and they had an average career ANY/A of 3.61 (+/- 3.3), including the eight players who didn’t qualify for having an ANY/A, usually by not ever throwing a pass in the NFL. As a point of comparison, Chase Stuart reports that the league-wide average in 2019 was 6.16. The average “median” indexed passer rating for the Parcels players was 57.7, but one standard deviation was +/-46.7. In other words, these guys were all over the place.

The remaining 80 players had an average career ANY/A of 3.26 (+/-2.7) and an average “median” indexed passer rating of 48.8 (+/- 44.8). So, that’s a little worse, I guess? Of course, these numbers show so much variance and have such wide margin errors that they are meaningless. This big lump of players doesn’t take into account draft position, the opportunity for playing time, or anything else. Still and all, it’s a big lump of players, and they were largely bad whether or not they met Parcells’ checklist.

Maybe the test found some duds, but did it at least identify the gems? No, it did not. Of the 21 Pro Bowl quarterbacks, only eight pass the Parcells test. Of the 17 quarterbacks with a median indexed passer rating was at or above 100 (basically, whose middle-of-the-road year was at least middle-of-the-road or better), only 8 passed the Parcells test. Finally, of the 24 quarterbacks whose career ANY/A was above 5.81 (set, as a Bears fan, at the level of Jay Cutler’s career performance), only 11 of them passed the Parcells test. With every objective measure I could readily check, the Parcells test succeeded in finding fewer than half of all “quality” quarterbacks.

Narrowing the Focus

Okay, so the Parcells test doesn’t work for the entire field. What about first-rounders? It is possible, even likely, that Parcells meant for his checklist to be applied only after strong quarterback contenders were identified. Still, this doesn’t help his case.

Of the 30 quarterbacks drafted in the first round, five pass the Parcells test (Andrew Luck, Baker Mayfield, Carson Wentz, Ryan Tannehill, and Tim Tebow). On the other hand, Matthew Stafford, Jameis Winston, Cam Newton, Sam Darnold, Mark Sanchez, Josh Allen, Jake Locker, Josh Freeman, and Johnny Manziel all have only half (or fewer) of Parcells desired traits, and Mitchell Trubisky, Patrick Mahomes, Baine Gabbert, Josh Rosen, Paxton Lynch, and Lamar Jackson all lack at least two traits.

The first list (the one Parcells would have preferred) has an average career ANY/A of 5.58. That’s not bad; it’s not good, but it’s not terrible. The second list (the one Parcells would have found questionable) has an average career ANY/A of 5.79. That’s better; it’s almost as good as Jay Cutler, and when a group fights its way back to mediocre while still including Josh Rosen and Paxton Lynch, it’s saying something.

What about late-round picks? Well, sure. We can salvage a very, very weak win for the Parcells test if we look only at late-round picks. If we look at quarterbacks taken in Round 4 and later, then the Parcells quarterbacks have a mighty 2.43 career ANY/A, driven mostly by the anomaly that is Ryan Nassib’s 11.42 ANY/A (who despite taking only 25 snaps, still has a 90% completion rating and a touchdown to show for his 10 pass attempts) and the legitimately valuable contributions made by Kirk Cousins and Dak Prescott. Meanwhile, the players with four or fewer of the Parcells traits are “led” by Trevor Siemian and are dragged down by Sean Renfree’s almost inconceivable -5.67 career ANY/A in order to average a 1.35 ANY/A overall. So—yes, we can manipulate the numbers to give Parcells a win.

Fellow Windy City Gridiron contributor Patti Curl made the very good point that with numbers like these, it’s sometimes helpful to use a metric called the odds ratio. I shy away from it because it can sometimes suggest a correlation without knowing causation. Still, it’s worth pointing out that for quarterbacks taken in Round 4 and later, 5 of the 19 who met the Parcells test had at least one mark of success (a pro bowl nod, an above-Cutler ANY/A, or an above-average median passer rating). Meanwhile, only 1 of the 41 others did so. So in this case, a late-round quarterback who met the criteria was more than ten times as likely to have some marker of success.

Even though it’s really hard to prove causation in a case like this, that’s a notable result. Overall, it’s easy to assume here that it might be worth making sure a player has an adequate number of starts and a decent completion percentage if you plan on burning a late-round pick on a backup. However, overall just not a useful checklist.

Where’s the Problem

Why is Parcells wrong? It seems like a reasonable list, yet there seems to be no way to make it work. First, it’s entirely possible that Parcells might have once had a point. However, these days, there is actually a negative correlation between someone playing as a senior and his career ANY/A (-0.25) or median indexed passer rating (-0.30). This makes perfect sense, because a lot of players who are good enough to make it as a starting quarterback in the NFL are going to come out early in order to start making money and avoid being injured while playing as “amateurs”, while other players are going to have a personal reason on some level to stay in school. The results still aren’t a noteworthy trend, but it isn’t surprising that being a senior is not always a good thing.

Likewise, quarterback wins aren’t a real stat. Football is the ultimate team sport, and there should be no greater need to invalidate the value of college quarterback wins that to point out that while Patrick Mahomes had only 13 wins in college over 3 seasons, Blake Bortles had nearly twice as many wins (25) in the same amount of time.

Ultimately, the Parcells list values things that are not signs of being a good quarterback. How many starts a quarterback has in college can be a sign of season length or injury as much as a sign of ability. It is also strangely redundant with valuing a number of seasons as a starter.

Final Thoughts

It’s worth pointing out that during the years when he served as a football executive for the New York Jets, Parcells drafted Chuck Clements in the 6th round of the 1997 draft (Clements had a sub-60% passing record and almost a 1:1 touchdown-to-interception ratio). Then in 2000 he drafted Chad Pennington, who basically matched his paradigm to a T and went on to have a solid career. Parcells later ran football operations for the Dolphins, during which time they drafted Pat White and Chad Henne, both in the second round, both of whom would pass his test (unless Henne’s 59.7% completion percentage just doesn’t cut it for purists). White would go on to record a negative career ANY/A in his limited career and Henne never managed to be more than a below-average journeyman.

In short, the Parcells test didn’t even really work for Parcells. Go back to the time that one time it “worked”, and Parcells drafted Chad Pennington instead of a number of other quarterbacks. Those other quarterbacks included Tom Brady—who failed the test on the basis game starts, seasons as a starter, career wins, and touchdown-to-interception ratio while in college.

So, there’s that.

Be sure you check out Josh’s previous entries in his Scrambling for a Quarterback series.