Hooks is doing a great job managing this blog, and in trying to make it more consistent and informative, he suggested that I provide one article a week, something of a "What GoPens! thinks" series. I thought it was a great idea, and so this will be the first installment. I don't yet have a catchy title for the series, but fortunately that's not a very serious problem.
Today I'm going to talk about situational metrics in the context of the Penguins defensemen this season. All stats are courtesy of NHL.com or behind the net.Line matching is a very unique feature of hockey. It is simply the idea that coaches work to get certain lines out in certain situations. They might have a "shutdown" line they throw out against the opposition's best players, or they might have a "power line" that composes the team's best skaters who get to play in the most favorable settings. The fact that coaches have the ability and insight to align certain players like this means that it is always important in hockey to figure out what type of situation players are typically used in. For me, my Holy Trinity of situational metrics is ice time, quality of competition, and zone start.
Ice time has an intuitive appeal to it -- the better players on a team are going to get more ice time. A quick glance at the top 5 TOI players this year includes Duncan Keith, Brian Campbell, Ryan Suter, Dan Girardi, and Shea Weber. Unsurprisingly, these guys are known for being some of the better defensemen in the league, and their coaches have verified that sentiment by relying on them so heavily.
For the Penguins defensemen, I want to look at total TOI per game, and even strength (ES) TOI per game. To focus only on those players with a big enough sample, I'm going to look at Paul Martin, Zybnek Michalek, Kris Letang, Brooks Orpik, Ben Lovejoy, Matt Niskanen, and Deryk Engelland. The results (players are ranked according to Tot TOI/game):
|Tot TOI/game||ES TOI/game|
There's nothing too surprising about these results. Someone who has never watched a Penguins game this year could look only at these numbers and easily figure out that Letang, Orpik, Martin, and Michalek are the team's top 4 defensemen. All four players are north of 21 min/game, whereas the next closest player barely has more than 18 min/game. Moreover, ES TOI pretty closely tracks Tot TOI. The Big 4 all play more than 17 ES minutes per game, whereas Engelland, the next closest, doesn't usually get more than 15 min per game at ES.
So far, it looks like the Big 4 are getting the tough minutes from Bylsma based on TOI numbers. But fortunately we have other stats that we can use to augment this picture and try to get an even more accurate idea of who is playing the hard minutes. The next statistic I'll use is quality of competition, specifically Corsi Rel QoC. This statistic is nothing more than a proxy for how difficult the usual opponents are for each player; hence the name quality of competition (QoC). The Corsi part means that the measure used to evaluate quality is Corsi, which looks at the shots +/- of a player while he is on the ice. Finally, the Rel means that it looks at these numbers relative to one's teammates, which provides a nice ranking system to figure out who is drawing the toughest assignments. Those players with a higher Corsi Rel QoC are drawing tough assignments, whereas those with lower (or even negative) numbers are playing weak players. The results:
|Corsi Rel QoC|
Once again, nothing surprising here. The Big 4 clearly lead the way in this department, as they're all hovering around a Corsi Rel of about 0.8. There ultimately isn't a whole lot of differentiation at the top. The big point is the drop off between Michalek and Engelland. Whereas Michalek is close to 0.8, Engelland is near 0.1, and both Niskanen and Lovejoy are playing even easier competition, indicated by their negative Corsi Rel.
So what we have so far is a pretty clear story. The Penguins' coaching staff has a ton of faith in the Big 4, as they routinely get the most minutes with the toughest assignments. Indeed, their assignments appear to be interchangeable, as the Penguins don't have a clear, far and away top D pair; all four defensemen alternate in the task of playing the opponent's top line.
The last stat we'll use to supplement our picture of what situation players are in is called zone start. In hockey, the ice is obviously divided into three zones, and coaches get to send players out to each zone based on what attributes they have. A strong faceoff guy with solid defensive skills might get sent out to take defensive zone draws more often, whereas dynamic playmakers and offensive juggernauts probably will have lots of offensive zone starts. The zone start stat attempts to capture this distribution by taking a player's total offensive zone starts, and dividing that by the sum of the player's offensive and defensive zone starts.
So, for example, if a player has 100 offensive zone starts and 50 defensive zone starts, his zone start will be (100/(100+50)) = 67%. Since this player has many more offensive zone starts than defensive zone starts, it is reflected in his offensive zone start % by the very high number. Thus, one can see that the higher the zone start, the more offensive zone starts a player has relative to defensive zone starts. The Canucks are notorious for taking this theory to the extreme: the Sedin brothers have an offensive zone start of nearly 80%, while Manny Malhotra has a paltry offensive zone start of 12%.
The Penguins don't engage in similarly extreme behavior, but they do allocate zone starts among defensemen according to ability. Based on what we've seen so far, we'd expect the Big 4 to have a lower zone start than the other defensemen. This is because the coaches trust them more, and thus give them the harder minutes in the defensive zone. Also, the opposing team usually gets their top line out in the offensive zone (which is the Penguins defensive zone), so the Big 4 are likely to be out in the defensive zone defending against top opponents more often. The results:
|Off. Zone Start %|
Voila, what do ya know. Lovejoy, Niskanen, and Engelland all have pretty high zone starts, which reflect the fact that they're usually not out against the other's teams top competition. Lovejoy particularly must be a liability in his own end according to Bylsma. On the flip side, we see that the Big 4 continue their dominance of tough ice time. Because the Penguins are such a good possession team, there aren't a whole lot of defensive starts to go around, which is why all seven of our regular defensemen have zone starts north of 50%. But those with the lowest totals are still the ones that we'd expect -- Martin, Oripik, Letang, and Michalek.
All in all, the story isn't that complicated. The Big 4 eat the big minutes, and the combination of Niskanen, Lovejoy, and Engelland take up the smaller roles that are leftover. With these situational metrics in mind, I'd like to talk about a few of the argumens regarding new defensemen for the Penguins that were thrown around recently. One was that trading for Hal Gill at the trade deadline would have been a very sensible decision. Those who said so didn't specify whether the Penguins would have to also give up one of their Big 4 dmen, but nevertheless, I'd like to look at Gill's situational metrics to see what he would have brought to the table. The results:
|ES TOI/game||Corel Rel QoC||Off. Zone Start %|
These numbers illustrate that Gill would have been a bad addition to the Penguins. Even on a bad team like Montreal this year, Gill wasn't getting top minutes. His roughly 13 min of ES ice time per game would have made him second to last on the Penguins, and he was only on the ice against relatively weak competition. To his credit, he was dealing with a rather low zone start. Yet the reason for that is most likely because Montreal isn't a great possession team, and thus has way more defensive zone starts to go around. One might argue Gill could have been useful because even though he was getting easy minutes, he was crushing it while he was on the ice. But that isn't the case -- Gill's on-ice Corsi is -9.49.
Thus, the story we're left with in terms of Hall Gill is that he is player getting easy minutes but simultaneously getting spanked by the competition. Even if you don't trust advanced stats, both the coaches in Nashville and Montreal were not giving Gill a lot of minutes overall, which means they didn't have a lot of faith in the quality of his play. Given that the Penguins' Big 4 already skates much tougher minutes and comes out way on top in terms of possession, replacing any one of them with Gill would have made little sense given his situational metrics this season. Moreover, the bottom guys like Lovejoy, Niskanen, and Engelland, though skating easy minutes, still had much better on-ice Corsi scores than Hal Gill, which means Gill would have been no better than any of them. Maybe Hal Gill is a nice 5th or 6th defenseman for teams who need help at that position, but the Penguins certainly didn't need him, and they especially didn't need his $2.25 million dollar contract that comes with him.
The second player I'd like to talk about is Simone Despres. Some have been recently clammering for the team to make a move and bring Despres up to the NHL permanently. In evaluating this claim, lets first look at what type of situations Despres has been placed in this year:
|GP||ES TOI/game||Corsi Rel QoC||Off. Zone Start %|
Of course, this is a very small sample, so it's clear we cannot infer a whole lot from this data set. But what we can see is that while Despres has been on the NHL team, he's been given very easy minutes. He plays little ES ice time (even less ES ice time than when Lovejoy plays) and does so against very bad competition. Additionally, he is played much more often in the offensive zone than the defensive zone. To Despres's credit, he has very good possession numbers on the ice. But that isn't terribly surprising: Niskanen and Lovejoy have very similar possession metrics, which is evidence that these numbers are being driven by the relative ease of the minutes.
So given that Despres is doing nothing different than what our bottom tier defensemen are already doing, the question now is whether he'd be a better replacement for one of the defensemen currently in the Big 4. Admittedly, we have little data on this point, and it's impossible to say with any respectable degree of certainty that Despres will or will not be able to handle tough minutes and still come out on top. Indeed, we don't know how he'd fare with those minutes because he's only played in a limited number of situations while he's been with the team thus far.
This does not mean that Despres will never have a place on the Penguins; as time progresses and he gets called up again, the coaches might flirt with giving him bigger minutes. If he gets tougher minutes (which we'll see in his situational metrics), and if he is able to still come out on top in the possession department, then the argument for moving Despres up permanently becomes a lot stronger. But until then, bringing Despres up permanently adds little to this team.
I'd let him continue to develop in the AHL, and focus on creating a role for him on the team next year. Both Lovejoy and Engelland are signed to league minimum contracts through the 2012-2013 season, but Niskanen will become an RFA this summer. If the team is serious about incorporating Despres, they might think about either letting Niskanen go to create room for Simone, or trading Lovejoy and Engelland to create some space. I think it would be a very bad decision for the team to do anything with the Big 4 on the idea that Despres will move up and take their spot. He might well be able to sometime in the future, but until we see some evidence of that, its best to stick with what we have.