Comparing Expected Pick Value with Expected Trade Value

I’m trying to re frame draft evaluations on the marginal value when compared to expected value of the pick position. There’s a surprising lack of public information on what the value of a given pick is, likely because it’s hard to pin down to an exact value. The two more commonly known charts I found from some googling are Jimmy Johnson’s Trade Value Chart and Over The Cap’s Fitzgerald-Spielberger Chart. I don’t know how good these are at projecting real-world trade value, but based on my comparison to player career value, they translate horribly as a measure of the base value of a pick. I also found this source, which seems to run a more sophisticated version of the baseline pick value than I’ve done here.

One large factor, is the one-number metric I use (stolen from Pro Football Reference) for valuing players struggles with distinguishing volume stat players from highly valuable players. Here’s a cherry-picked example of its limitations: my criteria would rank Blake Bortles 4-year value at AV = 42 compared to Andrew Luck at AV = 48, meaning Bortles was ~88% as valuable of a pick as Andrew Luck. There’s also perfectly rational factors, that result in a difference between trade value and pick value. Teams draft in large part based on need (hence why Packers fans were so upset with the Jordan Love pick), not on a commonly agreed upon value stat. So the value of a pick fluctuates from draft to draft based on how much competition there is for the players at the given pick range. However, I’m of the opinion (I’m trying to come up with a way to demonstrate this) that an at least equally large factor is that GM’s vastly overrate their ability to project talent.

This data is taken from NFL Drafts between the year 2010–2015.

Mean Approximate Value at the half-round marks

I’m forced to bin the values as the mean of every sixteen picks to smooth over the enormous variance pick-by-pick since I only used data from a combined six draft classes. I have some ideas on how to translate this mean into a linear decrease from bin to bin.

In both observed career value and trade value, there’s a substantial drop in pick value every pick you go down.

This grouped bar-chart took me way too long to create, but does a good job of showing how late round picks might be undervalued. Each group of bars represents half a round of draft picks.

Takeaways:

  • Aggressively trade-down from the top of the first round. That’s the section in which there’s the greatest disconnect between trade value and Approximate Career Value. The discrepancies between expected value and trade value are even more extreme, had I gone more granular than half rounds per
  • Given that we now expect that the drop-off in player value is far less steep than common wisdom dictates, volume is king. Honestly, just having a large rookie class is a virtue in and of itself, owing to their artificially constrained contracts. Even when they perform exactly to expectations, they outperform in value/$ when compared to FA acquisitions or players receiving a contract extension.

Data Scientist