Opinion Pieces dehaan5

Published on August 20th, 2014 | by Carey Haber


Puck Moving Defensemen

Last week, we took a look back at how the Islanders could benefit from a strong goaltending tandem. As we examined, one key aspect to any potential Isles’ success this year is their possession play. As we found last week, their goaltending can get them to a certain point, but it is possible that the real driver of how we can measure potential success comes down to controlling play. Or as the Isles’ themselves would say, “playing the way we need to play to win games.”

For the purposes of this discussion, we are going to refer to individual player Corsi. Corsi For is the amount of attempted shots (on net, blocked, or missed) that a team takes towards their opponent’s net. Corsi Against is the amount of attempted shots an opponent takes towards your net. Throughout this article, we will use Corsi For as a percentage, as to normalize the metrics to a scale of 0% to 100%.

Based on the definition of Corsi, and the implications that Corsi (and Fenwick) can have in terms of your expected win percentage, we can start to infer aspects of the game itself that can help lead to a higher “Corsi For” percentage.

One of these key facets is creation of offense from a team’s defense. We know the game moves faster than it did 15-20 years ago, and we know that mobile, puck moving defensemen have been at a premium on the free agent market for years. A big reason for these trends is success teams see when they can puPhoenix Coyotes v New York Islanderssh offense from the blueline. In 2012-13, the Islanders incorporated a defense of two key, veteran puck moving defensemen – Mark Streit and Lubomir Visnovsky – and were 19-12-4 with both in the lineup (5-5-3 when only Streit played).

After that season, one of Garth Snow’s major errors of the 2013 off-season was not replacing Streit with a capable puck mover, instead going with a rookie in Matt Donovan. While Donovan showed some very strong possession signs (53.5% CF), he was both unlucky in terms of goals against, and struggled with systematic and physical elements of the NHL game.

Coupled with Visnovsky’s concussion issues, the Islanders were left without a key, reliable, big minute driver of puck possession until Calvin de Haan was brought up to the team. The below chart shows the Isles’ Corsi when Visnovsky and de Haan play, versus when they don’t:


Source: ExtraSkater

As we can see in this chart, when Visnovsky and de Haan both played last year, the Islanders had very strong results. When only one of the two players played, the team struggled with possession. Surprisingly, when neither played, the team again had very strong results. This leads us to Thomas Hickey. The below chart outlines Hickey’s possession performance by segment:


Source: ExtraSkater

As we can see here, Hickey raised his performance pretty greatly in games where Visnovsky and de Haan were absent. Although the Isles lost some offensive power (as we will see below), they were able to keep the puck in the opposing zone.

Let’s now look at the first chart including the shooting percentages and save percentages of each segment:


Source: ExtraSkater + Hockey-Reference

In this chart, we actually see an inverse relationship to what we know is true. In other words, the Islanders see worse records when either both Visnovsky/de Haan play or when neither play versus when just one of them play. If we were to take this a step further, we can hone in on the save percentages and shooting percentages of each of these segments.

What we can see is that the Isles struggled in games that Visnovsky and de Haan both played due to an unsustainably low save percentage, which includes ten starts from Evgeni Nabokov and five from Anders Nilsson. In short, the goaltending seemed to eliminate any effects of strong possession the Isles put together last season.

Next, we have to look at the shooting percentages. If we were to look at these segments, we see very unsustainably high shooting percentages for the Islanders in games where only one of de Haan or Visnovsky played. Surprisingly, the goaltending was adequate in games when the Isles had a low Corsi For percentage, and coupled with an unsustainably high shooting percentage, the team (on the back of John Tavares, Thomas Vanek, and Kyle Okposo) performed quite well.

What we can see, however, is that the Isles’ shooting percentage when neither de Haan nor Visnovsky were in the lineup is uncharacteristically low. Without an offensive minded puck moving defenseman, it becomes infinitely more difficult to create offense through breakouts (a major issue for the Isles in 2011-12).

So, now that we know some of the causes of the Isles’ game results in each segment, we have to ask what contributing factors occurred that led the Isles to have a high Corsi during11.28.13_deHaan the period that de Haan/Visnovsky played, but without much offense attached to it. That brings us back to Thomas Hickey, who is a strong possession player, but is not offensively skilled enough to carry an offense by himself.

In short, although we saw some pretty interesting, major outliers from last season within the defined segments, we can certainly assume that for this season, the Isles are simply a better possession team and a better offensive team when de Haan and Visnovsky are in the lineup together.

Moreover, we can also expect more normalized numbers (in other words, a save percentage greater than .871) in a world where both players are able to stay healthy for elongated periods of time. Additionally, both players have pretty solid complements ready and able to step in either as partners or as complementary pieces to the up-tempo game in Travis Hamonic and Griffin Reinhart.

The bottom line here is that with normalized numbers and a solid contingency option in Thomas Hickey, the Isles’ fortunes on defense have the potential to reverse themselves from last season. Unfortunately, the team falls in short in depth, which underscores two things: the overall need for puck-moving defensive depth and the overall importance of Calvin de Haan and Lubomir Visnovsky to this team’s defense.

About the Author

7 Responses to Puck Moving Defensemen

  1. nbw says:

    Depth – hopefully TJ Bennen & Donovan can provide

  2. NDRE says:

    Donovan is interesting. There were evident holes in his game this past season, yet his possession numbers were downright stellar for a rookie defenseman (even for a defenseman, period). I’d love to see a post along these lines about Donovan’s WOWY numbers…. His combined Corsi On + Corsi Rel score was 13th last season (and de Haan 17th) among defensemen with a positive Corsi Rel QoC stat. (I linked in the charts, above, but basically the top-30 last season was Giordano, Vlasic, Muzzin, Timonen, Stralman, Keith, Seabrook, Doughty, Karlsson, Niskanen, Wisniewski, Brodie, Donovan, Hedman, DHamilton, MStaal, de Haan, Pietrangelo, Green, Demers, AGreene, Chara, Subban, Boychuk, BSmith, Campbell, Braun, Bouwmeester, Fayne…. Is there another defenseman on that list who is not top-4 quality? Were Donovan’s possession numbers fluky? Was he largely carried by de Haan? (This is why I’d like to see a WOWY analysis.)… To my eyes Donovan seemed to be a decent-to-average 5th defenseman last season, which is great for a rookie. But possession stats suggest he played like a solid top-4 EV defenseman, even if he was carried some by de Haan. (BTW, if Reinhart comes along nicely this season, I’d love to see Reinhart-Hamonic and de Haan-Donovan as dual shutdown pairs by season’s end. That would leave Hickey to carry the 3rd pair. (Whether Visnovsky stays healthy or not.)

    • Carey Haber says:

      Can take a look at Donovan. The question with him, even moreso than WOWY is whether he has the physical tools to keep up with play in the defensive zone, not whether he can push play forward. But, it’s something worth looking into. Stay tuned.

  3. Geoff M says:

    A few specific points

    1. It’s hard to gauge the effect of the two defensemen without supplying the team’s record during each period you cite.

    2. It seems to me that sv. % and sh. % are the variables most correlated with success, if we use PPG as a proxy for record. And with regards to sh%, are you claiming that defensive possession numbers are highly correlated with sh %? That may be true, but I would hypothesize that there are factors that have a larger role in determining sh %. John Tavares missing for a good portion of the season might have something to do with it.

    3. You only focus on 3 defensemen here, but the elephant in the room is Andrew McDonald. What was the Isles Corsi like when he was and wasn’t in the line up? What were the rest of the stats like?

    To my mind, the primary take home message from these stats and the subsequent analysis is that there’s a limit to what these stats can tell you, especially when you apply them to individual players (or even pairs of players). Analytics as applied to hockey is in its infancy – eventually someone will come up with some more informative stats that will provide a better picture of individual performance.

    • Carey Haber says:

      Thanks for the feedback!

      Surprisingly, Visnovsky and de Haan played the majority of their games together when Tavares was out – the games that neither were involved in were actually in the November-December swoon (of course). The points per game column in that chart show that the team played poorly when de Haan/Visnovsky were in together in terms of record, but that’s mostly due to the dreadful save percentage. It may not be random, but I think it is.. they just simply didn’t get the goaltending they needed last season, and both I think are quality overall players.

      But you are right, there are certainly some variables at play here that I did not touch on (MacDonald, Tavares’ injury etc.), which in some ways could have strengthened the argument. I think the overarching message, in terms of a predictive perspective, is that this team needs a healthy Visnovsky (very debatable if that is possible) and a healthy de Haan (we’ll see) to succeed, which was more of what I was trying to get at. That from a possession standpoint, it puts the Islanders in the best place to win if they can get better goaltending (which is probably likely). It would be very unlikely (though not impossible) for them to be a playoff team with a very below average possession rate.

      That all said, I agree with you that these stats are primitive in the grand scheme of the game. Most teams are already way more complex than just poaching things off the readily available sites. That’s something that will likely develop over a long period of time, so totally agree there. It doesn’t make these stats irrelevant (yet), but we do need to be wary of using them as gospel. On that I am in total agreement.

  4. Garik16 says:

    It’s pretty clear that de Haan and lubo are keys to the isles d corp being effective going forward but there are some issues with your analysis.

    First it’s better to look at corsi with the players on the ice than the teams corsi with those guys playing in games. Both de Haan and lubo had excellent corsi #s indicating they were both VERY positive possession players last year. So why was the team better off with none of them playing and worst overall with only one playing?

    Well you need to look at what else was going on. When both guys weren’t on the team the team was also for the most part without the “services” of Brian strait, who was the isles 2nd worst possession dman last year. As such the team was only playing one bad possession dman (amac) resulting in better possession.

    Strait was back by the time de Haan came up and moreover Matt Donovan (positive possession player) was sent down. Same thing here.

    By contrast take the hickey #s: why was he so had with one of them in the lineup? Well because that situation coincided mostly with his disastrous time with Andrew Macdonald where he had a sub 40 percent corsi. Again nothing to do with those guys being absent but just who he was paired with.

    Let’s also discuss shooting percentage: you argue that without a puck moving dman like de Haan or visnovsky, shooting percentages will be depressed. But if you look at the shooting percentage of the team when either guy is on the ice (http://stats.hockeyanalysis.com/ratings.php?disp=1&db=201314&sit=5v5&pos=defense&minutes=100&teamid=19&type=goals&sort=ShPct&sortdir=DESC) you see that there really isn’t much difference at all between isles dmen, and that dehaan falls below dmen like Brian strait- hardly a puck mover. Really the differences in shooting percentage of the team you noticed is more likely small sample size randomness.

    You’re ideas above are good and de Haan and lubo are key to this team. But you need to be careful: team results when a player does or does not play in the game can be caused by many many factors and are often misleading. A better tool for analysis is wrong how teammates play with and without a player on the ice in games he plays- you can find that at stats.hockeyanalysis.com. It’s not a perfect tool, but it is much better at proving what results are the effects of a player and what aren’t (if you want help using the site feel free to email me)

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Back to Top ↑