I wrote a blog yesterday regarding why I think defensive defensemen are overvalued in the National Hockey League. After posting, I received a lot of constructive criticism on the nature of my analysis – and rightfully so. I left out too many factors for the analysis to be legitimate. I’d like to remedy that. Today, I’ll be expanding on yesterday’s post to attempt to create an analysis that is fairer and shows the full picture. I’ll be attempting to demonstrate how defensive defensemen are not as effective as their offensive counterparts at reducing shots against and goals against. Despite their reputation of being more effective than offensive defenseman at doing so, I’m not convinced this is necessarily true.
Let’s start with changing the examples. In my last post I pitted stars against plugs. That’s not fair, as it shows bias in the analysis process. For this post, I’ll attempt to use more fair pairings of defensemen. Let’s take a look at the comparison between players such as Andrei Markov and Josh Gorges, Carl Gunnarsson and Jake Gardiner and Brooks Orpik and Matt Niskanen. These players are closer in time on ice and salary equivalents than the players I used in my previous post.
Additionally, I’ll be looking only at five on five shot totals that are adjusted to their zone start ratios to remove the way their coaches deploy them from interfering with the analysis.
Let’s start with Gorges and Markov. Markov is a very offensive defencemen while Gorges has impressed the Canadiens organisation with an aptitude for defending. Both are valuable, but does Gorges style of play really accomplish what it is believed to, that is reducing shots against?
When one adjusts their deployment – that is, removing the bias of offensive or defensive zone starts from the data – the differences in shot attempts is evident between these two players. Per 20 minutes of ice time that Josh Gorges is on, the Canadiens allow 14.03 unblocked shot attempts against and generate 12.705 for. Meanwhile, per 20 minutes of ice time that Markov is on the ice, the Canadiens allow 14.685 unblocked shot attempts against and generate 14.483.
It is important additional to this to mention how difficult their competition is. According to stats.hockeyanalysis.com — the same site in which I have received the information mentioned — Gorges and Markov have very similar competition. Based on the shot totals of their opposition, Markov has a fenwick quality of competition of -1.5 and Gorges has one of -1.7. Based on this, this information is unbiased in the competition they are facing and as it is zone start adjusted it is also unbiased based on the way their coach deploys them.
The rates tell an interesting story. When zone starts are adjusted, Markov allows 0.68 unblocked shot attempts against more per 20 minutes than Gorges does, but creates 1.7 shot attempts more. This means that Markov, per 20 minutes of ice time, creates a positive unblocked shot attempt differential of 1 when compared to Gorges rates. Gorges, however, does create marginally smaller unblocked shot attempts against with a 0.65 attempt differential. The unblocked shot differential individually comes to roughly -1.3 for Gorges and -0.2 for Markov.
Gorges is making 3.9 million this season and receiving 21 minutes of ice time including 57% of shorthanded minutes for the Canadiens; Markov is making 5.75 million and playing 24.5 minutes a night, including 74% of power play time. Salary information received from Capgeek.com, minute information from extraskater.com.
Let’s see if these results are consistent with some other examples across the league.
The next set of defensemen I will look at are from the Maple Leafs; Jake Gardiner and Carl Gunnarsson. Again, I will be using zone start adjusted stats provided from stats.hockeyanalysis.com to avoid the bias of starting too frequently in the defensive or offensive zone. Gardiner has a fenwick for — fenwick being unblocked shot attempts — per 20 of 13.552 and a fenwick against of 16.945. Gunnarson, meanwhile, has a fenwick for per 20 of 12.793 and a fenwick against per 20 of 17.321. Let’s take a look at that visually.
This case is much more telling than the previous. Gardiner generates approximately 0.8 unblocked shot attempts per 20 minutes of ice time more for his team that Gunnarsson. Meanwhile. he actually limits shots against more effectively than Gunnarsson; specifically, Gardiner allows roughly 0.4 unblocked shot attempts less against his team that Gunnarsson. Gardiners style of play is obvious to any that watch; while Gunnarsson is much more of a stay at home defensemen — not that he has not shown the ability to be a very effective puck mover — Gardiner is often pinching and looking to turn the play into an offensive one. For this reason, Leafs coach Randy Carlyle looks to defensive minded players like Gunnarsson in situations where he feels he needs defense first, despite this information demonstrating that offensive minded defenseman such as Gardiner can be equally if not more effective at defending as they have the ability to create more plays in the opponent’s zone. You can’t get scored on while you have the puck, after all.
Gardiner creates an on-ice unblocked shot differential per 20 minutes of ice time of -3.393 and Gunnarson creates one of -4.528. Gardiner and Gunnarson have very similar quality of teammates and competition as well. Stats.hockeyanalysis.com has Gunnarson’s teammates at a Fenwick adjusted rating of -31.8 and a competition of fenwick adjusted -1. Gardiner, meanwhile, has a teammate rating of -34 and a competition rating of -1. Gardiner is playing with worse linemates but against easier competition than Gunnarsson. These differences are not very significant and should not dilute the information provided in any way.
Gunnarsson is playing 19.5 minutes a night, including 54% of the leafs shorthanded time on ice and bringing in a salary of 3.45 million. Gardiner is playing 20.4 minutes a night including 42% of the Leafs power play ice time and making 1.2 million on an entry level contract, so essentially making the max that he can be at this age.
I’ve decided to take a look at a couple of Pittsburgh defenders next in Brooks Orpik and Matt Niskanen. Orpik has a reputation as a shutdown player, a reputation strong enough to earn him a spot on the U.S. olympic team. Meanwhile, Niskanen is a solid two way player who has become known more for his offense after enjoying a breakout year this season. Ask the average Penguins fan who is more effective at reducing goals against and I’m sure the answer would be Orpik; this is further demonstrated by the fact that Bylsma has such a love for the aging defender. Let’s see how this belief holds up in the shot attempt against data.
Orpik has a fenwick for per 20 minutes of 13.325 and a fenwick against per 20 of 13.840. Niskanen has a fenwick for per 20 of 14.624 and a fenwick against of 12.611.
From this we can see that Niskanen is superior in both creating shots for and reducing shots against. No one should be surprised by the former statement, but the fact that he is more effective than Brooks Orpik at reducing shots against will be surprising to many hockey fans. The reality of the situation is that Orpik’s playing style of being very stay at home is not necessarily as effective at reducing goals against as many believe, including coach Dan Bylsma.
The total differentials are substantial. Per 20 minutes of ice time, Niskanen creates a differential of roughly 2 additional shot attempts for the Penguins; Orpik has a differential of negative 0.5.
Niskanen plays 21 minutes of ice time per game including 51% of power play time and makes a salary of 2.5 mil. Orpik also plays an identical 21 minutes and is payed 3.75 million.
As for their competition and team mates, Orpik has a fenwick quality of teammates rating of 2.4 — very good teammates — and a competition rating of -2. Niskanen has a teammate rating of 1.3 and a competition rating of -1.5. From these we gain that both are playing with very good teammates, with Orpik having better teammates. On the competition side, both are very even with Niskanen playing slightly harder competition. Again, these facts do not heavily change the data, but if they did it would be slightly in Niskanen’s favor.
I think it would be interesting to look at the shot totals of all of the Penguins regular defencemen. Dan Bylsma really loves defensive defensemen, so it will be interesting to see how effective they are compared to the offensive ones employed by the Pittsburgh organisation. For defensive defensemen I have slotted in Rob Scuderi, Brooks Orpik and Paul Martin — I was reluctant to include Martin in that bin as he is not the typical defensive defensemen as he is incredible two-way but had to include him for symmetry as well as simply that his reputation and use in the Penguins franchise seems to be defense first, as demonstrated by the fact he starts 55% of the time in the defensive zone via extraskater.com. Their average fenwick for per 20 minutes 13.013 is and their fenwick against per 20 is 13.812. For the the offensive defensemen I will use Olli Maatta, Kris Letang and Matt Niskanen. Their combined fenwick for per 20 is 14.41 and their fenwick against per 20 is 13.461. Again, this is all zone start adjusted information from stats.hockeyanalysis.com, so it is not influenced by Bylsma’s use of these players. Let’s look at that graphically.
The defensemen with more offensive reputations come up on top in both accounts, but the surprising part is the shots against. It’s telling that these defensive defencemen in Orpik, Martin and Scuderi — all of whom have very good reputations for reducing shots and therefore goals against — are not as effective at doing so as the offensive counterparts on their team. The three defensive examples have a combined fenwick rated quality of teammates of 5.317 and a combined quality of competition of -6.9. Their offensive counterparts have a combined fenwick rated quality of teammates of 5.8 and a combined quality of competition of -6.4. From this we see that the defensive players are playing with slightly worse teammates but also playing slightly worse competition by nearly the same margin, so it shouldn’t change the findings.
Let’s take a look at the combined salaries and ice time of these players. The defensive players mentioned are bringing in 12.5 mil in salary and playing an average of 21.3 minutes a night. The offensive defencemen are making a total of 6.7 mil. However, these numbers are very unfair as Letang’s large contract does not kick in until next season and Maatta is on an entry level deal; they are both worth much more than that. They are getting an average ice time of 21.
It is not unreasonable to assert based on this evidence that the offensive defencemen on Pittsburgh are doing a better job of reducing goals against than their defensive counterparts and are doing a significantly better job of creating shots for, the latter of which is not surprising, but the former is. Additionally, they are doing so while apparently being valued equally based on ice time by Bylsma.
Based on these results and the results of the entire case study, I would assert that the reputation that many defense-oriented d-men hold as having abilities to reduce shots and therefore goals against more effectively than offensively talented defensemen seems to be false. Although this is a relatively small case study, it is still telling. More work can be done in this topic and I hope to in the future; however, as for evaluating the effectiveness of the average defensive defensemen, this information leads me to believe that they are viewed as preventing more shots and goals than they in fact do. Just because someone is defense oriented does not necessarily mean they are actually going to prevent more shots against a offensively oriented one. Fans and analysts often look at a defense-first d-men and assume that he keeps shots against down better than an offensive defense because of the nature of their play, even when this may not be true. It doesn’t matter the manner that a player is preventing shots against; it only matters whether they do. For this reason, I assert that many defensive defensemen are not actually more effective than offensive ones at reducing shots and goals against, even though they may hold that reputation.
All zone start adjusted and fenwick based quality of competition and teammates received from stats.hockeyanalysis.com. All other stats received from extraskater.com. Salary information retrieved from CapGeek.com.
Thanks for reading!
Follow me on Twitter @LukaRyder