Tuesday, June 25, 2013

Drafting By Position

The following is a re-post from a while back but with the draft coming up, I think it's worth re-visiting:

About a week ago, I was talking with a friend about 2013 eligible draft prospect Darnell Nurse.  He's huge and athletic and while I'm not a pro-scout, I'd guess that he's a guy with an awfully high ceiling.  I suggested that some team that barely missed the playoffs was going to be awfully lucky to draft him and that I wouldn't even be surprised if he snuck into the top-10 on draft day.  It was at that point that my friend suggested that taking a defenseman in the top-10 was a risky proposition.

We shot some names back and forth of forwards and defense who had worked out, where star players came from, Cam Barker, &c. &c. Are your odds better of getting a good player in the first round if you take a forward?  At what point, if any, are defensemen generally safer picks?  We couldn't really come to any reasonable conclusions without doing the work.

So with that, I set about putting together a spreadsheet with 10 years of draft data (1999 to 2008) where each pick in the first three rounds was categorized as an All-Star, an Impact Player, a Replaceable Player, or a Bust.  For now, I've held fast to defining an All-Star strictly (must have actually been an All-Star) but I do think this analysis would benefit from a little more flexibility in this regard (ex. Hamonic and Subban are not All-Stars but Justin Williams and Alexander Frolov are.)  An Impact Player is a top-6 forward or a top-4 defenseman, a Replaceable Player would be a guy who has predominantly been an NHLer but generally in a support role, and a Bust would be a guy who didn't have a significant NHL career. 

The following charts represent the percentage of players who fall into the defined categories per player selected at those positions.  In terms of raw numbers, roughly two forwards are drafted for every one defenseman.

The first round results were a little surprising.  The blue segment represents players who fell in the All-Star range and Impact range combined.  While the bust rate among defensemen selected was slightly higher than it was for forwards, the number of impact defensemen per defenseman taken actually outpaced forwards in the first round.  Now again, given that this is done in percentage terms, there are more impact forwards taken than impact defensemen, but when you adjust the numbers to a 'per defenseman taken' and 'per forward taken', you see that a greater percentage of selected defensemen seem to provide meaningful contributions to their teams.

The second round also sees a greater share of defensemen providing impact-level performance.  This was a little less surprising based on the guesses that I made heading into the exercise.  Take note of how high the bust rate is in the 2nd round (over 75% overall.)  

Again, take note of the bust rate. Also, there's a greater proportion of those who 'make it' who are replacement level here.  Pretty tough to find a top-6 or top-4 guy by the time you get to the 3rd round.  Once again, we see that the numbers seem to slightly favour defensemen.

So what does this data mean?  Well, first of all, in isolation it suggests that not enough defensemen are being taken in the first or second rounds.  If the number of defensemen selected in early rounds were to increase, we would likely see these numbers come together a bit.  Secondly, it could mean that GMs believe that drafting an impact forward is more valuable than an impact defenseman.  While I'm pretty comfortable, generally speaking, with the categories I employed, what they don't take into account is the order of magnitude that these players impact their teams -- as far as these numbers are concerned, Malkin and Frolov are the same guy.

What I am comfortable saying is that I prefer that my team take a defenseman over a forward in the second round.  The number of cases where we see ultra-high impact players selected in the second round is limited and in instances where it occurs, those players generally seem to be defensemen.  Not only this, but the numbers pretty clearly suggest that your odds of getting any kind of asset are better if you take a defenseman in the second round rather than a forward.

Moving forward, it may be interesting to break apart the first round into pick-segments and depending on how well received this post is, I may take that on.  For the time being though, it's safe to say that I wouldn't be avoiding a guy like Darnell Nurse with a pick in range of tenth overall, certainly not on the basis of the position that he plays.

1 comment:

Anonymous said...

Love to see this about goalies through the whole draft, since it seems like a crapshoot but likely that's confirmation bias.