A Deeper Dive into NFL Draft Picks

With the NFL Trade Deadline approaching, it is important for teams to level-set on expectations for the draft picks they give up or receive in return

Last week, I called upon NFL teams to be more active at the Trade Deadline. The Philadelphia Eagles then read my blog post and immediately traded for Safety Kevin Byard 😊. Trade activity has been minimal since then, but teams are reportedly working the phones in advance of tomorrow’s deadline. As teams decide what draft pick compensation to offer in exchange for veterans to bolster their playoff chances, I thought it was important to dive into what to expect from said picks.

To do so, I evaluated data from the 2000-2019 NFL Drafts. I chose 2019 as the cutoff, as these players are now largely on their second contracts, giving us enough time to evaluate their careers. For each pick in the draft, I analyzed the Weighted Approximate Value (wAV) through the 2022 season of each player taken during this window at that selection. The goal of this, was to generate a 25th, 50th, 75th, and 90th percentile outcome for each selection in the draft using wAV scores. To simplify things, I then attempted to bucket the draft to see what kind of output teams can expect from certain parts of the draft, taking an average of the wAV scores for all the picks in each bucket. The bucketing is a bit imprecise due to compensation picks changing where each pick falls in the draft every year, but can also easily be modified in the Excel file as desired. Lastly, because wAV scores are a little abstract, I tried to find an example of a player with each specific wAV score. The example player was chosen by finding the most recent drafted player with said wAV score. For example, Deandre Baker was drafted in 2019 and was the most recent player drafted with a wAV score of 6. Because wAV scores grow larger throughout a career, some of the players with higher wAV scores were drafted earlier. As an illustration of this point, Julio Jones was the most recent player drafted to achieve a wAV score of 100 and he was taken in 2011. The results are below:

The important thing to note is that the example players were picked based on wAV score alone, not where they were actually drafted. A 25th percentile outcome at the 1st pick in the draft would be to select someone with a wAV score of 50. Buffalo Bills Linebacker Matt Milano is an example of such a player, but he was taken in the 5th Round of the 2017 NFL Draft - not with the top selection. What this serves to illustrate is that a 25th percentile outcome for someone picking atop the draft would be to take someone that has had a career like Milano. wAV scores are not a perfect measure of career success, and players like Running Back Kareem Hunt certainly benefit from having played longer than others like EDGE Rashan Gary - who had only one full season as a starter going into the 2023 season and had a wAV score of 15 going into the season. Guard Earl Watford is an example of a player who had that same wAV score but across a longer career, showing some of the limitations of this metric as any credible football fan knows that Gary the better player of the two. Still, as I wrote about last week, there is no one metric that can tell us how good a player is, so we need to do our best with the data that we have and think critically about it.

A few things stand out from this analysis. Firstly, by the time the fourth round of draft rolls around, most players do not develop into much more than a backup in the league for a few years. The median pick at the start of the fourth round achieves a wAV score of 8. An example of a player with that score is Offensive Tackle Greg Little who has bounced around the league since being drafted in the second round in 2019, and is currently a Free Agent. Secondly, while median outcomes tend to drop as we get later in the draft, there are a few buckets where they largely flatline. There is almost no difference in outcomes between picks 6-10 and picks 11-20. Outcomes between the late first (picks 21-32) and early second (picks 33-42) are similar as well. Still, as a whole, median outcomes drop off far more rapidly throughout the first and second rounds than in the rest of the draft. Finally, there remains significant downside after the first few picks of the second round. Throughout the middle and late second round, a 25th percentile outcome would be to draft a player with a wAV score of 9 - an example of which is Defensive End L.J. Collier who has barely played outside the 2020 season. A visualization of some of these trends is below:

The point of this analysis is not to say that teams should abandon the draft. Drafting well is an essential part of building a contender under the constraints of the salary cap. Teams should be very careful when deciding to deal their picks in the first and early second rounds because of how much outcomes drop off after this stage of the draft. Contenders, however, should feel relatively confident in dealing picks even in the third round if they believe that the player they acquire will make a meaningful impact on their championship run. Chances are, said player will be more productive than whoever is taken with the draft pick. Furthermore, these teams will have the ability to replenish the cupboard through trading down from the late first into the early second, where there is limited drop-off in outcomes, picking up some more mid-round picks in the process. Contending teams also often lose more free agents than they sign, meaning they’ll receive compensatory picks in future drafts as well. Both these options help further de-risk trading for veterans if their cap situation allows.

For non-contenders, this helps show how much of a crapshoot much of the draft is. Therefore, the best way to increase the likelihood of success in the draft is to accumulate picks, which can be accomplished by trading impending free agents that they do not expect to be on the team next year. Having more darts to throw on draft day makes it more likely that a team will achieve a higher percentile outcome with one of those picks. This, I’d argue, will help them more down the line than holding onto a player they are likely to lose in Free Agency.

I am happy to share the Excel workbook I used with anyone who requests it. Please don’t hesitate to reach out via email at gauravv@mit.edu.