Draft Viz: Some interesting looks of the draft

Updated: January 11, 2018 at 3:12 am by Garret Hohl

The draft is one of the most interesting places to investigate in hockey. You have predictive analytics, growth curves, game theory, opportunity costs, and many other areas of study interacting.

Sometimes I find information interesting, anecdotal or otherwise, but doesn’t really pertain relatively speaking to what I’m focusing on. I recently fiddled with the draft vizualization centre (over at hockeyprospectus.com by Matt Pfeffer) and found some interesting trends.

I thought I’d share these trends:


The graphs above is every draft selections combined GVT value for the drafts between 1985 and 2011.

There have been many excellent studies on perceived pick value, built by using how certain picks have been traded for others historically. I have not really seen enough to satisfy my curiosity though on how efficient the draft truly is (although there is a lot out there).

It is remarkable how efficient scouts are in selecting the top 30 picks overall in the first two graphs. This efficiency though drops significantly when looking past the top 30. Scouts also are not nearly as efficient with defenders or goaltenders.

There are some issues with the vizuals limited to overall value and not per number of picks. When selecting all positions, this is not an issue; however, it is when looking exclusively at one position where there may be more or less players selected at that spot.

Defenders being less in number than forwards for drafts – and goaltenders even less – do allow for outliers to hold a larger impact than with forwards. Any draft analysis expanding on these trends may want to look at distribution and use median instead of the mean for averages.

For those interested, here are forwards for picks 0-30 and picks 30-60: