Pages

Wednesday, May 7, 2014

Draft performance by position


Yesterday’s post looking at the drop-off in value from the first to the second player off the board at each position was a bit of a tease. I fixed the overreliance on tables with a Tableau visualization (find it here) and now it’s time to make the cross-position comparisons begged by the data.

To start with, I want to change Approximate Value. The distribution built into the stat is the allocation of Draft Value to each position according to the Jimmy Johnson chart. Keeping the total points of AV the same, I’m going to redistribute by salary instead. From my database of salaries 2001 to 2012 I will pull the “starters” at each position. This means I’ll take the top 32 QB salaries, the top 64 Guard salaries, the top 128 DB salaries etc – resulting in the distribution shown below:



This makes a lot of sense to me. You hear a lot of talk about Russell Wilson’s extremely low salary helping the Seahawks allocate money elsewhere. This is absolutely right. With elite QBs routinely pulling down ~$18-20 million he is a steal at under $1 million.

You do not hear nearly as much talk about Luke Kuechly being a steal for the Panthers. He is, of course, a steal for the Panthers, but his making $3.4 million against the cap in 2014 while elite LBs top out at James Lauranaitis’ $10 million and no one else exceeds $7.5 million.

Given that we can pull together an expected value for the first four years in terms of Adjusted AV, which looks a lot like this:



Applying that back to every pick we can now compare how well positions do relative to the expectations for each pick that actually took place. That’s complicated, but at least the visual is simple. 


Quarterbacks show up as a good choice, because they deliver value that’s worth more in the marketplace. TEs adjust downward significantly, as expected. WRs and DBs, the two positions identified yesterday as having top picks that fail to outperform the next player off the board at their position, bounce all over the place as player value is hard to predict. Defensive Ends have an interesting trajectory with early picks underperforming, but players in the second round being well above expected value. Feel free to play with this and see what other patterns are (or might be) there.

No comments:

Post a Comment