104 Comments

Data Returns: Statistically Profiling the Ideal Quarterback

| February 12th, 2017

This is the 2nd installment of a monthly offseason piece I’ll be doing here at DaBearsBlog, helping fill the content void of the long offseason. Each one will be a numbers-crunching look at something Bears related in which I attempt to earn the “Data” moniker so kindly bestowed on me by the comments section regulars and, more importantly, answer a Bears question that I’ve been wondering about. If you have anything you’d like me to look into, let me know in the comments or email me at woodjohnathan1@gmail.com and I’ll see what I can do. 


By all accounts, it seems the Bears will be acquiring the man they hope will be their quarterback of the future this offseason. Ryan Pace was spotted scouting pretty much all of the top quarterbacks in person throughout last fall, and his end of the season press conference was centered around a discussion of what he’ll be looking for in a franchise quarterback.

With that in mind, it would be wise for any Bears fan to pay close attention to the quarterbacks at the top of the draft this year. I started doing just that back in November, when I looked at quarterbacks drafted between 2011 and 2015 and found teams looking for a starter should focus on the top of round 1 or round 2 (http://bit.ly/2lhS3t0). Luckily for the Bears that fits either of their first two picks.

Building an Ideal QB Profile

Now I want to focus on what they should be looking for with one of those picks (thanks to DBB’s Andrew Dannehy for giving me this idea). Here’s how I went about doing that:

  • I looked at all 1st and 2nd round QBs drafted between 2011 and 2015 and compiled a bunch of data about their physical measurements, passing stats from their last year in college, and team success in college. The full list can be seen here: http://bit.ly/2kQ8v2L.
  • I split the QBs into guys who are established starters (Newton, Luck, Mariota, Winston, Tannehill, Bridgewater, Dalton, Carr), guys who might be starters going forward (Kaepernick, Garoppolo, Bortles), and everybody else.
  • I averaged the data together for each group and especially compared starters vs. everybody else (non-starters). 6 traits were identified that were significantly different.
  • For each trait, I sorted the quarterbacks from best to worst and looked for a “benchmark” value, which most of the starters hit and most of the non-starters missed. This always fell such that 5 or 6 of the 8 starters were above the benchmark; there was typically a significant dropoff after this point such that this was a logical cutoff.

Based on this, here’s the ideal profile I found to look for in a highly drafted QB coming out of college:

  • He should win at least 77% of his college starts (6/8 starters hit, 3/9 nonstarters)
  • He should win a conference title (6/8 starters hit, 4/9 nonstarters)
  • His final college season should feature at least 8.7 yards per passing attempt (5/8 starters hit, 3/9 nonstarters)
  • His final college season should feature a touchdown on at least 7.3% of his throws (6/8 starters hit, 3/9 nonstarters)
  • His final college season should feature a TD/INT ratio of at least 3.7:1 (6/8 starters hit, 2/9 nonstarters)
  • His final college season should feature a college passer rating of at least 166 (5/8 starters hit, 2/9 nonstarters)

There didn’t seem to be any difference in the physical profiles of the QBs based on their height, weight, or hand size at the Combine. The important part of the Combine for QBs is their interviews, but we don’t get that data. Ignore the measurables; they are basically irrelevant for QBs.

Putting the Profile To the Test

I then went back and looked at every QB in the study to see how many of the six benchmarks they met (starters shown in bold).

  • 6/6: Newton, Mariota, Bridgewater, Dalton
  • 5/6: Luck, Garropolo
  • 4/6: RG3, Manziel
  • 3/6: Carr, Bortles, Manuel, Smith
  • 2/6: Winston, Weeden
  • 1/6: Kaepernick, Ponder
  • 0/6: Tannehill, Osweiler, Gabbert, Locker

Every single player who hit at least 5 benchmarks is now an established starter except for Jimmy Garoppolo, who has spent 3 years sitting behind Tom Brady but is likely to be traded this offseason and get his chance to start next year. He certainly seems to fit the profile of a successful NFL QB based on his performance in college, though it’s worth noting that he’s the only QB on this list who played in FCS and thus comparing his stats to the others’ might not be fair.

As you move to players who hit fewer of the benchmarks, the odds start to drop. There are certainly some successful players there, including Carr, Winston, and Tannehill, but the busts outnumber them and all fall in this region.

Winston and Tannehill are two interesting cases. Winston hit both team markers but none of the individual success thresholds, though he would have hit all of them in his first season as a starter. Tannehill hit none of them, but his situation is fairly unique as somebody who switched from WR to QB halfway through his college career. A few other players on this list have had moderate success-Kaepernick, Bortles, and RG3 immediately come to mind, but none have established themselves as the long-term answer at QB that the Bears are looking for this offseason.

This list also highlights the shortcomings of this approach. RG3 and Manziel both looked like solid prospects based on hitting 4 of the 6 benchmarks, but both washed out in the NFL. They both had off-field issues that contributed significantly to those problems, and both played in offenses that are not typically transferable to the NFL. I am not evaluating personality or style of offense in this approach at all.

Evaluating 2017 QBs

Now that I have a general profile for what recently successful QBs have looked like coming out of college, let’s examine the top options in the 2017 draft to see if any of them fit the bill. The results for the 5 QBs who I’ve seen generally mentioned as potential round 1-2 prospects are listed below.

  1. DeShaun Watson: 2/6 (winning %, conference title)
  2. Mitch Trubisky: 1/6 (TD/INT ratio)
  3. Pat Mahomes: 1/6 (TD/INT ratio)
  4. Brad Kaaya: 1/6 (TD/INT ratio)
  5. DeShone Kizer: 0/6

Well that’s not very promising. Based on this approach, the QBs at the top of the draft this year do not look like good prospects. The only one who fits any sort of profile matching a recently successful individual is DeShaun Watson, who matches the same 2 team-focused traits that Jameis Winston did. Unlike Winston, however, Watson does not have a previous season on his resume that would have checked off many of the individual performance benchmarks.

Of course, it is worth noting the difference between correlation and causation here. When looking at a sample size of 20 QBs, I identified 6 traits to separate 8 of them from the other 12. This really separated 5 of those 8 out, but there is no guarantee that will be a predictive model going forward. It might just be that, in this particular sample, these successful players all coincidentally shared those traits. The smaller a sample is, the greater the likelihood of that correlation being random. 

I could look farther back to try and increase the sample size, but I don’t want to do that because of how much the game has changed for passing offenses in college and the NFL. Passing stats from 5 years ago are difficult to compare to now, but 10 years ago would be completely obsolete.

Conclusions

Let me caution again that this is only intended to be one way of evaluating prospects and is certainly not going to be infallible. Even within this method, the success rate is not 100%, and this includes zero film study on my part, so it’s not intended to be all-inclusive.

With that said, this approach certainly dictates staying away from the QBs at the top of the draft, with a possible exception of DeShaun Watson. It also makes trading for Jimmy Garoppolo look like an appealing option. Comparable college profiles to Garoppolo have a very strong success rate in this study, though of course they all played in the FBS while he was in the FCS. 

Before looking into this, I was firmly in the “no” camp on Garropolo, as I don’t think New England would trade a QB they thought was good when Tom Brady is approaching 40 years old. But this study has me re-considering that stance. Consider me officially on the Jimmy G bandwagon. 

Tagged: , ,

© Da' Bears Blog