The Effects of Draft Position on PER

 by Jon Nichols

            One of the things I like to do when performing my statistical studies on the draft is to use an expected PER (Player Efficiency Rating, developed by’s John Hollinger, check it out here: for each player.  This way, we can determine how good or bad a pick was based not only on performance but also on expectations and salary. 

            However, the expected numbers I use are somewhat arbitrary.  I decided that a #1 pick had to have a PER of 16.8 or better to be considered a success.  I then subtracted .2 from each following pick to come up with a list of expected PER’s.  I did look at players’ PER’s to come up with those numbers, but they weren’t scientifically chosen. 

            I decided to figure out exactly how good a player you should expect based on how high they were picked in the draft.  To do this, I took a look at the first 30 picks from every draft from 1980-2000.  I then calculated the median PER for each spot in the draft.  I chose the median instead of the mean because I wanted to account for the players who never appeared in many games but also not have their extremely low or high PER’s affect the study too much.  Once I got all the median scores, I made a scatter plot that looks like this:


            For those who aren’t statistical experts, I’ll explain those equations above.  First of all, the y= equation has the number -0.2451 in it.  Basically, for every subsequent pick in the draft, we can expect the player’s PER to drop by .2451 points.  For example, we can expect the #10 pick in the draft to have a PER 2.45 (.2451 x 10) points higher than the #20 pick in the draft.  Also, the graph shows an R-squared value of .8542.  This is a very high R-squared value, indicating the line does a pretty good job of estimating median PER’s for each pick. 

            As I mentioned earlier, I used to subtract .2 from each subsequent pick.  In actuality, I should have subtracted .2451.  This means my earlier study was a bit too harsh on the low picks.  Also, #1 picks do not fit to the line very well.  Top picks in the draft tend to be far and away better than everyone else, so we should expect a bit more from them.

Finally, in any draft, the draft picks will end up being much different than the average.  The 10th pick could be terrible, and the 20th pick could be an all-star.  What this study does is confirm the common sense that higher picks do better and it sets a baseline for how much we should expect from a player based on pick number. 

Information from and was used in this study.





Copyright © 2009