Measuring Teams’ Draft Success Using Analytics

Your ads will be inserted here by

Easy AdSense.

Please go to the plugin admin page to paste your ad code.

I had started a project last year where I attempted to analyze and rate various facets of different NHL management groups. And it goes without saying that running the draft is one of, if not the most, important aspects of any general manager’s job. So, naturally, I wanted a method to effectively measure just how well (or poor) these teams were doing at building through the draft. It will also come as no surprise that I took an analytics-focused approach to
doing this.


A lot of work has been done to try and evaluate teams at the draft. Much of this work involves a lot of subjective analysis though, often assigning “grades” to players, one at a time. I wanted to take a different approach.
Building off of what I started in a post last summer, here (so excuse my plagiarizing myself), I started by looking at the average GVT/year for all players who had been drafted in the last 21 years (excluding this previous year’s draft). For those of you who are unfamiliar with GVT, Hockey Prospectus explains it best:

Goals Versus Threshold. Developed by Tom Awad
of Hockey Prospectus, GVT measures a player’s
worth in comparison to a typical fringe NHL player.
GVT has two major advantages over most metrics: it’s
measured in goals, which are easily equated to wins,
and it is capable of comparing players across multiple
positions and multiple eras.

To put it in other words, GVT is a measure of a player’s value in terms of the amount of goals he generates/prevents for his team. It’s obviously not going to be a perfect indicator of a player’s talent level, and it has been known to undervalue defensive play, but it allows forwards, defencemen and goaltenders to all be evaluated by the same metric. That, and I trust GVT more than I trust some of the other, more subjective methods of evaluating players.

I also chose to only look at the average GVT/year for the first seven years of each player’s career in an attempt to evaluate each player in roughly the same window of his career. This way players like Ryan Smyth wouldn’t be penalized just because they chose to play out the very end of their careers. They’re still only being graded on how they performed when they broke into the league.
So after then calculating each player’s average 7-year average GVT/year (or GVT/yr(7) for short), I calculated the average GVT/yr(7) of all players drafted at each individual slot in the draft. I then plotted it all out:

GVT7_All Positions_All Years
Pretty much what you’d expect. Not a huge surprise there. To make it a little more presentable though, I started combining averages in draft slots of three after the 30th pick, which lead me to this:

GVT7_All Positions_All years_Smooth
Great. At this point three things caught my eye:

1) If you look closely, you can see the value for the 2nd overall pick is slightly higher than the 1st overall pick, definitely not something you’d expect. I’d like to think that Rick DiPietro is looking down on us from his cloud of money and medical gauze right now, reading this with a grin on his face,  nodding his head with approval.

2) Picks 11-14 have apparently been a sweet-spot for teams.

3) Like I had brought up in last summer’s post, there is hardly a difference in value between the 100th pick and the 200th pick. Looking at this and nothing else makes me really want to see a team try to trickle down from their 4th round pick, acquiring as many picks as they can along the way; not unlike what MacT did last year, trading the 37th pick for the 83rd, 88th, 94th, 96th and 113th. Is it a coincidence that the Oilers GM is also very involved with analytics and the draft?

Moving on to the fun stuff though, I followed up by fitting a curve to the data:

What a stunner. Not too shabby if I don’t say so myself. With this curve, I was now able to evaluate whole teams at the draft over great periods of time. That only required for each team taking each drafted player’s actual GVT/yr(7), subtracting his predicted GVT/yr(7), and then averaging out the differences. Completing that, I now present to you the results, starting with every pick from every draft round from 1992-2008 (2008 so that we don’t start digging into the recent draft years where only players from the top rounds have even played any games, skewing the results):

Your ads will be inserted here by

Easy AdSense.

Please go to the plugin admin page to paste your ad code.


So what this table tells us, looking at a team like Detroit as an example, is that on average, their draft picks from 1992-2008 produced an extra 0.2 goals/season more than what they were expected to, given their slot in the draft. That’s some exceptional drafting. On the other side of the coin, there sit the Oilers at a lowly -0.23 goals/season. At least they’re still better than the Flames though, right?

It’s a little harder to appreciate drafting success from so long ago, however, so not to take anything away from Colorado’s stellar drafting record over this time, I’m going to narrow our timeframe a bit:


Seven Points of Interest:

1) It was at this point that I knew that I was on to something, as teams that everyone has implicitly known to be poor drafting teams (Edmonton, Calgary, Columbus, Atlanta, Florida, etc.) were all found at the bottom of the pile.

2) San Jose’s management group doesn’t get enough credit.

3) For anyone who wants more ammunition against the Leafs, here you go. According to this, Toronto was one of the best drafting teams from 2000-2008. Why is that ammunition? Because that means that to consistently finish towards the bottom of the standings since that time, their team’s asset management must have been (and possibly continues to be) just horrific.

4) Just doing an eyeball test it looks as if the richer teams tend to group closer to the top while the poorer teams get grouped closer to the bottom. That makes me wonder how much successful scouting is dependent on actually having successful scouts, and how much of it is dependent on simply spending more than the next team.

5) Building off that, what would happen if Nashville had Toronto’s budget?

6) Edmonton still beats Calgary (or do they just suck less?).

7) Chicago and Boston, both powerhouses, both fairly pedestrian by this measure. Chicago must have had some big busts scattered in there to bring down what would be the boost that drafting Toews, Keith and Seabrook gets you.

In Closing

No method of measuring a team’s drafting ability will be perfect, far from it. I think there are enough positive signs from these results though that we can have some confidence that these numbers aren’t totally out to lunch. Moving forward, I’ll try to analyze some of these numbers in more detail, starting next post with a breakdown of the 2000-2008 drafts, position by position.

If you have some stones, try predicting which teams have been the best/worst at drafting forwards, defencemen and goaltenders.

This entry was posted in Uncategorized. Bookmark the permalink.

Leave a Reply