Further to the Issue of Age and QB Prospects

Jon Moore posted a thought provoking piece earlier where he wondered whether the age of a rookie QB should affect our expectations for what kind of prospect they might be.  Be sure to read Jon’s piece in full because this will be just a short addendum and there’s certainly more work to be done on this topic.

I went to the last 10 years of draft data and asked pretty much two questions to further my understanding of the issue.  The first question I asked was “Does draft position affect our expectations of QB quality?”  The second question I asked was “Assuming the answer to question #1 is ‘yes’, does age impact the probability that a prospect will exceed or underperform the draft pick expectation?”

First let’s look at the QB quality/draft pick issue.  Using PFR’s CarAV metric, I plotted about 12 years worth of QBs based on where they were drafted and their CarAV/G.  Here’s the graph I got.


I’ve shown the rsq on the graph so you can see that it’s .17.  NFL teams are doing a little better than throwing darts, but they have a lot of room to improve also.

After I had that trend line, I calculated each player’s CarAV over or under the expectation and then broke it up by the age that the players were as rookies.  I got this graph:


The trend is certainly going in the direction implied by Jon’s piece earlier today.  I think there’s a lot more to do on this, but I also think that as Jon laid out, there are a lot of reasons that we should expect that the relationship works like this.  Let’s think about them for a minute:

  1. NFL prospects have an option on turning pro, which is to say that they can do it when their value is at its highest, which is to say that the oldest players may have just run out of time.  They weren’t good enough to come out as underclassmen, so they stayed in school until they couldn’t stay any more.  So the older age cohort is likely also less talented.
  2. As Jon pointed out, they get to play that last college season against players that are often sometimes 1, 2 or 3 years younger than they are.  They might stand out for that reason and that could impact our perception of them.
  3. The place I might differ with Jon is in terms of applying the results of this analysis.  Instead of creating a threshold and then cutting things off at 23 or anywhere else, I might construct an age penalty in a model so that younger players get more credit for college performances and older players get less credit.
  4. The other thing that could be done is that you could create penalties in the college measures based on age, so that it’s possible to compare underclassmen with seniors.

As I said, I think there could be more done on this topic.  One thing that could improve the analysis would be to gather more than 10 years of data.  You could also experiment with using different metrics to assess QB results.  I used CarAV because I could get it quickly.  But in any event, I think Jon Moore was on to something and this could be a rich area for evaluating QBs in the future.

What we do


Sign-up today for our free Premium Email subscription!

© 2019 RotoViz. All rights Reserved.