The More the Merrier? Correlating Success with Quantity in the NFL Draft

As a long-suffering fan of the Cleveland Browns, I am no stranger to just about every NFL Draft strategy out there.  Trade up for a QB?  Been there.  Trade picks for players?  Yep.  How about players for picks?  Of course. 

For the past couple of years, the new-look, analytics-driven Browns have been amassing picks for the new "acquire assets" model championed by the Patriots.  But does it have basis in fact?  Let's dive in?

We will start with a simple Tableau scatter plot of win percentage vs. number of picks for the year.  What do we get?

Uninspiring to say the least.  The complete lack of a correlation is backed up by adding a regression line.

The lowly R-Squared value of 0.02% tells the tale.  Yeesh.

But what if we look at the teams a year or two in the future?  To do this, I created a delayed field by joining on a calculated field and two years to the draft season.   See below.

This join will show us win percentages correlated with number of picks from the prior year (2016 record vs. 2015 pick count).  The results are below.

Still nothing great, but the R-Squared percentage is up to 1.5%.  However, it actually shows a negative effect! 

What about 2 and three years down the road?  Those results are below.

As we can see, while the trend moves back to a positive one as we would expect, it is still not a significant predictor of success (R-Squared values of 2.1% for a two-year lag and 0.4% for a three-year lag).

Well, once again Browns fans' hopes are dashed.  Oh well.

Back next week with the conclusion of the NFL Draft series!

Author: Chris Bick