Photo by Lisa Gansky, via Wikimedia Commons; altered by author
While preparing statistics for a few upcoming posts on on-ice contributions, I decided to do a quick study on the share of on-ice shot attempts taken by defensemen versus forwards. The metric I’m using, which is a spin-off of an old one whose name doesn’t quite capture it right, is what I’m calling on-ice shooting proportion, or OSP. The results were quite interesting, and I decided that I should test the data a little further and see what we could find.
For this post, I focused on 5v5 ice time for two reasons: 1) it comprises the overwhelming bulk of on-ice play (about 80%), and 2) it’s less likely to be artificially boosted by something similar to a blueline-driven powerplay approach. The first graph that caught my eye focused on comparing the entire forward and defenseman populations, from 2007-08 through 2014-15:
That certainly looks like a trend to me. Generally speaking, the average defenseman contributed 16.2% of all on-ice shot attempts this year; keep in mind, in a 5v5 situation with an even distribution of OSP among all players, you’d be at 20%. I wanted to dig deeper, though, and see if there was any differences between top players and the bottom.
The uniformity in that trend is remarkable. Teams are either a) having their defensemen shoot more often, or b) choosing defensemen who tend to shoot more often. Either way, it’s generating a greater share of attempts from the point, and it’s coming from the top on down.
Not quite satisfied, I decided I wanted to test this hypothesis a little further, this time looking at assists. Taking on-ice goals-for, I looked at what proportion of those goals were either first or second-assisted by the forward and defenseman populations:
While second assists seem to vary (though overall slightly trending up for defensemen), the first assists are, again, pretty clearly on the move. You’re going to have to bear with me on the next one, because I again split the groups into top to bottom lines and pairings:
The second assist groupings, as you can see, are a real mixed bag, so much so I don’t even think the overall forward vs. defensemen trend is convincing. On the other hand, let’s look at those first assist groupings a bit closer:
While the defensive third pair is a bit up-and-down, the first and second are clearly on the move, in the opposite direction of the first and second forward lines. The third forward line is mixed, but the fourth forward line is pretty astonishing. Remember the third graph from my post on parity? In it, the zone starts for the top 3 forward lines seemed to only have slight trends, but the fourth showed a pretty drastic drop in zone starts over the years, suggesting teams have moved towards a sort of defense specialization for them. That’s certainly supported by their drop-off here.
Chalk this up to another potential impact of the pressure of parity, and the demand for control of the puck. Teams are gradually realizing that they need more-complete defensemen — ones that can move the puck, not merely stop it — and whether by strategy or selection, they’re getting it.
Hockey Graphs 
This would explain why league average shooting percentage is dropping slightly.
I wonder if this might be because teams are collapsing a bit more and are better at defending the blue. I wonder if individual dmen are shooting more often, or if there are new guys getting more shots.
I would guess the latter. I have an upcoming post where I demonstrate that, by this OSP metric, players tend to retain their share of the shooting individually.
Yea, just did a quick check for guys that played every season from 08/09 to 14/15 (107 guys). Average change was only +.55 percent from year 1 to 7 with
Histogram of changes:
Yea, and after looking into it more, you are definitely right. Interesting. Looking forwardto that post.
So, in keeper fantasy hockey pools, if the trend continues, defencemen may be of slightly increased value, and forwards of slightly decreased value?
If it continues like this, and I’m confident it will…yes. But moreso because it’ll close up a fantasy market inefficiency. For the last few years, there was a pretty substantial gap between reliable top scoring defensemen like Karlsson, Subban, and Weber, and the highly-volatile next tier of defensemen. That group was a near guarantee for 50-points, while the next tier varied so wildly your expectation was 15 points less – but the big difference was that top group was larger than the similar top forward group that would score 15 points more than the second tier. In those years, if you were picking past 6 or 7, you’d do just as well to go D, F, or even D, D if you had strong forward sleepers in mind. This year, for instance, I had a low 1st round pick in a league and went Karlsson and Weber (Subban was picked in-between), then made later-round picks of Hornquist, Vrbata, and Tarasenko. Nothing lost in my forward group, and a reliable 15 point boost.
With more offensive-inclined defensemen, this could close the gap between the defense tiers.
I should say, 30-point predictive boost, since I netted both Karlsson and Weber – which ended up around 25-point because Weber performed below expectation. Still kept me in the top of the league for the whole season.
Ok, but what’s the upshot of this? Definitely an interesting trend, but what on-ice effect does it have? Scoring is trending downward league-wide, but it would be interesting to break this data down on a team-by-team basis to see if some teams are driving this defense-heavy shot approach more than others, and whether those teams are generating more goals or improving their overall goal differentials. There would be some noise to filter out in team-by-team statistics, since personnel changes can herald a change in offensive approach, but I think it would help us understand the effects of what we’re seeing above.
Other questions: We know that shooting percentages are lower for defensemen; is this trend dragging down shooting percentages overall? Are defensive shooting percentages climbing because defensemen are more involved in the offense, and are taking more shots from higher-percentage areas on the ice? Are the defensemen stealing shots from the forwards, or are teams that give defensemen the green light to fire away simply generating more shots?
No worries, I plan on following this up with exactly the kinds of tests you’re referring to 🙂
In the meantime, I did look at this idea of defense share of shooting historically, and noted its poor correlation to overall team performance: https://hockey-graphs.com/2014/02/18/nhl-defensemen-shooting-contributions-1967-68-to-present/ That said, it’s not quite the same metric at play in that post. Stay tuned…