The impact of offense only D-men

[Hagged]

There was a post on the Ducks board with a link to an article that caught my attention.

http://www.sportingnews.com/nhl/story/2014-05-12/nhl-zone-entry-stats-anaheim-ducks-la-kings-sami-vatanen-stanley-cup-playoffs?utm_source=dlvr.it&utm_medium=twitter

At first glance the impact of Vatanen to the Corsi of on ice team mates seemed too high to be true. Then I noticed Vatanen had a higher percentage of offensive zone starts and lesser QOC than most of his team mates. This would obviously skew the stats for the lower ZS% higher QOC team mates when their stats are compared with and without Vatanen. But how to correct for that zone start difference?

There is a sentence in the article "Vatanen was entering the offensive zone 59.5 percent of the time with control, which was nearly twice as successful as any of the other blueliners on the team (Lindholm was second with 37.3 percent, while most of the others were sub-20 percent). Only Cam Fowler was better at successfully exiting the defensive zone." What does this basically mean? Vatanen doesn't like to chip it in, or doesn't lose control while trying the zone entry / exit?

Vatanen was mostly spending time in the AHL due to his size and defensive shortcomings. Those who watched the Ducks last season saw that Vatanen brought a puck possession element to the third pairing when he was in the line-up, but he had problems without the puck in the D-zone going against Thornton for example.

So my question is this: How to rate the impact of offense only D-men on team performance through advanced stats without having to rely only on Corsi stats which obviously are high for these players as they generally are getting easier mach-ups with high OZS%.

Where do we see the size handicap of Vatanen in underlying defensive stats? If he's bad at defense, it should show up somewhere. How does his stats compare to other OFD in the NHL?

Click to read more...

 

[Hivemind]

Generally speaking, "offensive" and "puck moving" defensemen tend to put up much better Corsi/Fenwick/Shots-For numbers than their "shutdown," "defensive," or "stay-at-home" counterparts.

Here are some articles that touch on various points you brought up in your post.
http://www.mc79hockey.com/2014/07/orpik-niskanen-joe-morgan-and-the-engine-room/
http://www.broadstreethockey.com/2013/7/4/4487304/save-percentage-variability-regression-defense
http://www.hockeyprospectus.com/puck/article.php?articleid=1467

 

[Ho Borvat]

There was a post on the Ducks board with a link to an article that caught my attention.

So my question is this: How to rate the impact of offense only D-men on team performance through advanced stats without having to rely only on Corsi stats which obviously are high for these players as they generally are getting easier mach-ups with high OZS%.

Where do we see the size handicap of Vatanen in underlying defensive stats? If he's bad at defense, it should show up somewhere. How does his stats compare to other OFD in the NHL?

Here is an example I will bring up:

Lots of Canucks fans will say things like "Bieksa gets caught pinching way too much"/Always makes boneheaded plays/turnover machine etc.

The way you have to look at it is, does the offensive boost outweigh the occaisional lost puck battle/turnover/etc

Thomas Drance has discussed this before with Bieksa, that while he does have the odd mistake, he more than makes up for it by generating offense (to offset the occasional defensive lapse).

 

[Aela*]

You can look at things like On-Ice sv%, On-ice sh%, OZS%, OZF%, TOI, GP etc when looking at dmen who play on the same team.

Using Vatanen/ANA as an example, his OIS% is 92.1% (good for 6th among the team, discounting Alex Grant because he only played 2 games) while his OISH% is 5th among the team, again discounting Grant (and also Yonkman) at 10.05%.

Now he only played 48 games, so these numbers might fluctuate, and you should take them with a grain of salt as he's only played 1 real season with the club.

From what you can see with those advanced stats however, he was more of an offensive/2 way dman during the season who got a bit above average ofz starts, but also finished in the ofz more than avg as well (3rd on team).

All in all, he did pretty well during the season especially for a rookie. Now fast forward to the playoffs:

His numbers go a fair bit down, he has the worst OIS% on the team at .889, but he also has the 2nd highest OISH% on the team at exactly 10%. He also split time with Fistric and other players I believe being the #6 dman as he only played 5 games in the playoffs, and had the 2nd highest OZS on the team behind Fistric.

More than anything else his OISH% show that he's more offensive than defensive IMO as they stayed fairly static while his OIS% went way down (and to be fair, so did his PDO).

So tl;dr: OIS%/OISH%, TOI, GP, etc etc

 

[Hivemind]

I wouldn't use On-Ice sv% or On-Ice sh% to evaluate defensemen. I already linked one of these in post #2, but I'll throw it up here again.
http://www.broadstreethockey.com/2013/7/4/4487304/save-percentage-variability-regression-defense
http://www.broadstreethockey.com/2013/6/16/4430864/shooting-percentage-regression-goals

 

[NikF]

does the offensive boost outweigh the occaisional lost puck battle/turnover/etc

I think what sort of gets discounted within that framework is the concept of stability. Sure over the long run you might be just fine having the offensive boost outweigh the gaffes, the problem I have with that is that a wildly volatile game (constantly create chances, constantly give up chances) is in my mind too risky in a playoff format. I don't know much about advanced stats but instinctively I would think a team that has a more flat performance curve has less inherent risk of ending on the wrong side of the curve in one of their playoff series, than a team that gives up a lot and creates a lot does. I would use the same logic individually on offensive defensemen that while net positive in the long run have overall shoddy defensive game or regular gaffes in form of scoring-chances against. I don't know if that makes any sense, but I feel like there is a factor of stability that gets discounted when someone brings up the "well if his offensive boost outweighs whatever he gives up, then it doesn't matter" argument.

 

[Ho Borvat]

I think what sort of gets discounted within that framework is the concept of stability. Sure over the long run you might be just fine having the offensive boost outweigh the gaffes, the problem I have with that is that a wildly volatile game (constantly create chances, constantly give up chances) is in my mind too risky in a playoff format. I don't know much about advanced stats but instinctively I would think a team that has a more flat performance curve has less inherent risk of ending on the wrong side of the curve in one of their playoff series, than a team that gives up a lot and creates a lot does. I would use the same logic individually on offensive defensemen that while net positive in the long run have overall shoddy defensive game or regular gaffes in form of scoring-chances against. I don't know if that makes any sense, but I feel like there is a factor of stability that gets discounted when someone brings up the "well if his offensive boost outweighs whatever he gives up, then it doesn't matter" argument.

I think a lot of this is explained by score-effects.

If you're up 3-1/4-1 in a game, d-men aren't likely to pinch and make risky plays. You typically see less shot generation by the leading team because they are more concerned with minimizing the other teams opportunities.

Vice versa, the trailing team will have their d-men jumping up more/pinching and generally creating more offensive chances.

Again, I use Vancouver as an example, but if you're defending a lead , you are likely feeding more minutes to Hamhuis/Tanev than you are Bieksa/Edler. And you are likely asking Edler/Bieksa to sit back and play more conservative.

I think a lot of this is more circumstantial than anything.

 

[KaseMeOutside]

Sami vatanen belongs in the NHL, the idea that the ducks have possibly blocked him by signing clayton stoner makes me SICK!

the most overrated dmen in hockey are "stay at home" "tough" dmen who can't move the puck, they may have one use and thats killing penalties.

was vatanen sheltered? yes he was, but he also excelled in that role and he helps a team more than playing bryan allen and clayton stoner, like it's laughable he could be blocked by them

 

[Hivemind]

I think what sort of gets discounted within that framework is the concept of stability. Sure over the long run you might be just fine having the offensive boost outweigh the gaffes, the problem I have with that is that a wildly volatile game (constantly create chances, constantly give up chances) is in my mind too risky in a playoff format. I don't know much about advanced stats but instinctively I would think a team that has a more flat performance curve has less inherent risk of ending on the wrong side of the curve in one of their playoff series, than a team that gives up a lot and creates a lot does. I would use the same logic individually on offensive defensemen that while net positive in the long run have overall shoddy defensive game or regular gaffes in form of scoring-chances against. I don't know if that makes any sense, but I feel like there is a factor of stability that gets discounted when someone brings up the "well if his offensive boost outweighs whatever he gives up, then it doesn't matter" argument.

I see it in reverse.

For the sake of this discussion, let's accept the premise that "offensive defensemen" both create and allow more chances than "defensive defensemen." In other words, they play higher event hockey. The more events that take place, the more stable a probablistic system should be. Low event hockey will be more volatile from a probablistic standpoint, as there will be fewer events to help the sh%/sv% regress towards their true talent levels. In other words, if you minimize the events that occur, the greater the odds that one atypical event (a blown save or odd bounce leading to a goal, for instance) can influence the outcome of a game/series. Granted, if your team is outgunned in terms of possession, sh%, and/or sv%, minimizing events may be a wise choice. That's essentially how Montreal managed to top Washington and Pittsburgh in 2010.

 

[NikF]

I see it in reverse.

For the sake of this discussion, let's accept the premise that "offensive defensemen" both create and allow more chances than "defensive defensemen." In other words, they play higher event hockey. The more events that take place, the more stable a probablistic system should be. Low event hockey will be more volatile from a probablistic standpoint, as there will be fewer events to help the sh%/sv% regress towards their true talent levels. In other words, if you minimize the events that occur, the greater the odds that one atypical event (a blown save or odd bounce leading to a goal, for instance) can influence the outcome of a game/series. Granted, if your team is outgunned in terms of possession, sh%, and/or sv%, minimizing events may be a wise choice. That's essentially how Montreal managed to top Washington and Pittsburgh in 2010.

The way I choose to think about this is as follows. This is all hypothetical, but let's choose Doughty and Karlsson. Let's assign a numerical value to a performance. Let's say Doughty's performance curve is flatter than Karlsson's.

Let's assume this is a "typical" performance stretch for both:

Doughty
5, 6, 7, 5, 4, 4, 5, 6, 5, 6

Karlsson
5, 8, 3, 7, 5, 3, 3, 8, 8, 3

Over that time-frame - Doughty's total value contributed is 53. Karlsson's total value contributed is also 53. Now let's assume their teams are equally good. In a playoff format for a team to win the Cup they need a minimum of 16 games. It is possible that Karlsson contributes greatly but if we assume that his performance is inherently more volatile due to the type of game he plays then it is enough that he runs over a stretch of "3s" for a team to be eliminated. If Doughty underperforms and if we do believe that his performance curve is flatter then he is not hurting the overall chances of the team as much as Karlsson is. If I have it correctly we know that a player can under/over perform his talent for a stretch of time then I am concerned what happens with a player like Karlsson (in this example, I don't know for sure whether that is in fact a good example) when he is on the downward slope of the steeper performance curve (think of it as a sine-wave).

I guess you could say I'm worried what defensively suspect offensive defensemen DO when they're in a stretch where they are underperforming their true talent level. What was Mike Green doing when the puck didn't go in for him? Over the long-run, yes it doesn't matter if he's contributing more offensively. What happens when you're in playoffs, you're playing against elite teams, and your #1 pure offensive defenseman has a bad stretch? The puck isn't going in, but he's still running up and down and giving up defensive gaffes? There is no long-run to recover that back. A stretch of hypothetical "3s" might end your season. Where if a Doughty turns in a modest 4,5,4,4 there is enough bend but not break for the rest of the team to pick up the slack. The eye-test equivalent would be an unspectacular Doughty game he doesn't have much impact but he still doesn't get burned, an unspectacular Karlsson game he still runs up-and-down but he doesn't convert, his opposition does though (replace names with someone else if needed, not trying to make this into a Doughty vs Karlsson debate).

 

[Hivemind]

What makes you think that offensive defensemen are any more subject to game-to-game volatility in their performance?

If you go purely in terms of point production, there's obviously volatility there. But that doesn't mean their play actually deteriorates or that their teams as a whole perform worse. To use Mike Green as an example, the puck wasn't going in for him a whole lot this past season, but he still posted the best possession numbers on the Capitals. Only three Caps managed to post CF% over 50%, with Mike Green leading the pack. Orlov was one of the two others, and he spent a majority of his time as Green's partner and saw his numbers decline away from Green. Schmidt also did well with Green and saw his numbers decline away from Green.

To look at it from another perspective, what do defensemen with suspect puck management skills Do when they're in a stretch where they are underperforming their true talent level?

 

[NikF]

What makes you think that offensive defensemen are any more subject to game-to-game volatility in their performance?

Their job description. The premise that offensive output excuses defensive gaffes plays precisely on the fact that an offensive defenseman is excused giving up chances against if he converts on more chances for. That does work over the long run if we're talking about a quality offensive Dman. To me that's volatility in itself what else would giving up prime chances in order to have more expeditions offensively be other than a hedged bet that you will come out on top from risky situations? The volatility stems from the fact that a poor offensive showing leaves a cliche offensive defenseman with very little value provided. Similar to a poor defensive showing for a defensive-dman. If you're equating Mike Green with a defensive-defenseman then I'm with you. Defensive-D are already recognized as specialists and paid/valued accordingly less. However elite offensive defensemen whose entire worth is based on the premise that offensive output is greater than defensive gaffes should not be held in the same regard as elite two-way defensemen. If Erik Karlsson has a poor showing offensively he's toast compared to Zdeno Chara or Drew Doughty because their game doesn't hinge on the fact that they have defensive gaffes to be forgiven on as a regular occurrence.

 

[Hivemind]

Their job description. The premise that offensive output excuses defensive gaffes plays precisely on the fact that an offensive defenseman is excused giving up chances against if he converts on more chances for.

See this is where the disconnect is. The offensive defense isn't excuse for giving up chances against because he converts chances, but rather because he generates chances. A lower conversion for the defenseman rate over the short term, as is likely given the low incident rate of goals, does not equate to worse play, lower chance creation, or lesser offense from the team as a whole.