The Approximate Value of a Draft Class

The latest newsletter from KCSN Sports Data Scientist, Joseph Hefner

I got a request in the KCSN Discord to write about the expected value of a draft pick based on where they were drafted compared to the value we have seen them produce on the field. That is a really fun question, and one that is surprisingly difficult to answer with any certainty. Of course, that did not stop me from trying!

First, I want to say that I absolutely love it when people come to me with article ideas. Coming up with stuff to write about is hard, friends. When I first started this job, I had a dozen or more ideas I wanted to write about. That was three years and 128 articles ago. Please, if you have ideas you think are interesting, send them to me! Especially in the offseason, my schedule is very open.

Ok, back to draft value! The problem with comparing the expected value of a draft pick to the realized value is the definitions. Both expected value and realized value have to be defined. There are astonishingly few holistic stats that try to map out the value of a player compared to the rest of the NFL. Outside of second contract value, which has its own flaws, and needs 4+ years before it’s useful, there is little that is not arbitrary.

The only three stats that I am familiar with that give a player value rating normalized across the entire NFL are PFF WAR, Madden Ratings, and Approximate Value (AV). PFF WAR is not a public stat. I do not think a private individual could get that stat even with the highest tier subscription. If I did this analysis using Madden Ratings, I would likely face a riot. That only leaves AV, and that is the stat I chose to use for this.