Scrambling for a Quarterback: Observations on Later Picks

The Chicago Bears need a quarterback. While they have Nick Foles and Andy Dalton on the roster, Foles has clearly never regained the form he had in his magical run in Philly, and whatever Dalton might have been, he seems to be in the twilight of his career now. Much has been made of the notion that the Bears are picking too late in the draft to get a truly “good” quarterback, and that this means that Ryan Pace must trade for an established veteran, make a blockbuster trade to move into the top five (selling the farm along the way), or simply kick the can down the road to 2022.

Why? What if he simply drafts a quarterback at #20, #52, or (gasp) #83?

Right now, the Bears hold three picks between 20 and 100, and they have a lot of holes on the roster. What if by some strange wonder Ryan Pace doesn’t trade all over the place and simply makes picks at each of those spots, with one of them happening to be for a quarterback? Foolish, right? No more foolish than anything else Ryan Pace might do, and perhaps smarter than a lot of other choices.

The truism is that it takes at least three years to evaluate a draft, so it’s probably too soon to pass a verdict on the class of 2019 or 2020. However, over the ten years from 2009-2018, 28 different quarterbacks have been taken with picks #20-#100. That’s a respectable number of data points.

The Basics

I wanted to be stingy in my evaluation. I didn’t want a Brock Osweiler or Geno Smith sneaking into my tallies as a success due to a vague set of metrics. In order for a player to count as a “push”, he had to have at least 16 starts and at least a season qualifying as the starter (per Pro Football Reference). Anything less than this and I counted the player as a miss.

However, in order to count a player as an actual success, he had to have at least two seasons as the primary starter for his team, had to have at least 32 starts, and have to have a passer rating of 80 or higher. I also wanted at least two fourth-quarter comebacks or game-winning drives.

The results were surprising to me.

First, only 12 of the 28 were outright misses, but they were every bit as bad as a casual fan might guess: Pat White, Jimmy Clausen, Ryan Mallett, Matt Barkley, Johnny Manziel, Garrett Grayson, Sean Mannion, Paxton Lynch, Christian Hackenberg, Connor Cook, DeShone Kizer, and Davis Webb. Yikes. Woof. I’m a Bears fan and that’s a pretty disgusting list of quarterbacks (two of whom started for the Bears at some point, unfortunately). Worse, three years (2009, 2015, and 2017) only saw misses taken between #20 and #100. Not a single starter-level QB was taken in this range. These are the results that most Chicago fans have been afraid of.

More encouragingly, though, there were some “push” candidates. Players who in retrospect seem like they probably were worth trying out, even if things didn’t work out in the long run: Tim Tebow, Colt McCoy, Brandon Wheedon, Brock Osweiler, Geno Smith, Mike Glennon, Jimmy Garoppolo, and Mason Rudolph all fill in this group. These eight players are not going to earn a GM bragging rights when taken at #20, but they also seem like better uses of #83 than a rotational defensive tackle.

Most importantly, there were also eight outright hits (a rate of ~30%): Andy Dalton, Colin Kaepernick, Russell Wilson, Nick Foles, Teddy Bridgewater, Derek Carr, Jacoby Brissett, and Lamar Jackson are all in this group. Jacoby Brissett is the only one of these eight who hasn’t earned a Pro Bowl nod, and five have playoff wins. There was at least one success story found in five of the ten years. As a quick and efficient (if not thorough) cross-check to see if some of this success rate was because these players were more likely to go to strong teams, only two of them went to teams that had enjoyed a winning record the prior year (Brissett and Jackson).

Obviously there are no guarantees, but it is really encouraging for a team that is locked out of the top of the draft and still needs a passer.

A Strong Class Muddies Waters

One thing that struck me right away is that something interesting happens with the data when I rule out any year (hits and misses alike) where there weren’t at least three quarterbacks taken in the first round. This drops us down to 20 quarterbacks, but only seven of those are misses, and half of the “marginal” cases are removed. All eight of the hits remain (improving this to a 40% success rate). When there were a number of quarterbacks taken early, the failure rate of late-round quarterbacks decreased and the relative success rate increased. This merits repeating—the so-called failure rate of quarterbacks drafted out of the top of round one could be an artifact of weak quarterback classes as a whole.

There is a simple explanation that is no more than a possibility, but it is a possibility. Imagine that I’m the GM for a team that needs a quarterback in 2013. My top choices are E.J. Manuel, Geno Smith, and Mike Glennon. No matter how I look over the prospects, and no matter how much I go digging in the rough for diamonds, they just aren’t there to be found. In this case, I am fairly likely to spend a pick on the marginal prospect and hope for the best. Investing the draft pick doesn’t make the player better, though, and so I end up with busts and backups.

However, now imagine that I need a quarterback in a year like 2011 or 2012. In 2011, there were six quarterbacks taken in the first two rounds, and so Andy Dalton and Colin Kaepernick went back-to-back at the top of the second round. Likewise, in 2012, there were five quarterbacks taken ahead of Russell Wilson at #75. These years there were rich quarterback classes generating considerable background noise. GMs faced overwhelming pressure to get things right—even though they needed to parse through far more “high end” prospects than usual. Stronger candidates can slip, just as a matter of course.

In fact, of the twelve quarterbacks who were drafted after four or more other quarterbacks (they were #5 or later) and were drafted in this range, there are still three misses (Mallett, Cook, and Webb). However, half of the twelve were hits (Dalton, Kaepernick, Wilson, Foles, Brissett, and Jackson). This suggests heavily that in rich quarterback years, there is a decent chance of finding at least a starting-level quarterback simply by waiting out the field and letting a talented prospect fall (presuming, of course, that the team in question can identify that prospect once he has fallen).

Predicting Success

Bill Parcells thought that in order for a quarterback to be worthy of drafting, the player needed to have seven traits. The player needed to be a senior who started for at least three years (and across 30+ games) and who graduated. He also wanted the player to have a 2:1 touchdown-to-interception ratio and a 60% or better completion rate while winning at least 23 games. Two of these factors (whether or not the player graduated and the final win:loss rate) were not readily accessible reading over Sports-Reference.com, and the results even from the five traits that were available were so weak that it barely seemed worth checking. Thirteen players possessed all five of the other traits.

Four were successes, five were pushes, and four were misses. That’s not very useful, even if it is a touch better than the general population.

However, what is interesting is that there is a much simpler test available, and while I have not done a lot of work to validate it, it suggests that college football either rewards good players or at least produces good players. Imagine that I give every player credit for 10 bonus yards per start, and then draw a line at requiring 10,000 passing yards—I get twelve quarterbacks: Mason Rudolph, Sean Mannion, Colt McCoy, Derek Carr, Matt Barkley, Russell Wilson, Geno Smith, Andy Dalton, Colin Kaepernick, Nick Foles, Davis Webb, and Teddy Bridgewater. That’s six hits, three pushes, and three misses. If I disqualify quarterbacks with a sub-2:1 TD:Int ratio per Parcells, that only removes Mannion from the list (it only removes Mannion and Hackenberg from the total list, actually). Still, that’s a really useful tool to start a conversation about who might be worth drafting, if nothing else.

I am a little concerned about a rough-cut metric that excludes Lamar Jackson but includes Davis Webb. However, it at least suggests that Parcells’s actual test is vastly less useful than a number of alternatives. Is it helpful to know if a player in college had a lot of starts and completed his passes? Yes. However, there are a number of ways to get that information. It is actually more useful to know if the player in question has the raw number of successful pass attempts (and the resulting number of yards) to suggest that he is kept in the game by his coaches.

Conclusion

Ryan Pace does not have to make a dramatic move to find a starting quarterback (or even a Pro Bowl quarterback). Roughly every other year, a future starter falls out of the Top 20 and then makes an impact. More than that, if Pace has the chance to draft an accomplished college passer who fell simply because of a deep quarterback class, there is every reason to think that he should at least try.

Given the hit-or-miss nature of the NFL Draft, and his own track record evaluating quarterbacks, taking the best player available at #20 and then letting his scouts suggest which passer to take at #52 (or even #83) might in fact be a better strategy than putting all of his eggs in one basket and making a desperation move to try to get “the guy”.