By the Numbers: Optimizing the Hurricanes’ Defensive Unit

When it comes to the strengths of this Carolina Hurricanes team, the defensive unit looked to be where this team would excel. However, having good defensemen on the team is only half the battle. Those strong players will not be able to maximize their impact properly unless they are deployed and grouped in the most optimal ways.

Knowing how to pair a team’s defensemen can make or break that team’s season. After a disappointing 2014-2015 season in which the Los Angeles Kings missed the playoffs, head coach Darryl Sutter decided to balance out his defensive pairings more than they typically had in the past. Normally the Kings iced a top pairing of Jake Muzzin and Drew Doughty, a pairing which typically dominated the league, but left much to be desired out of the Kings’ other two units.

Last year, the Kings had Doughty with Brayden McNabb and Muzzin with Alec Martinez. Those two pairings were truly elite by every advanced metric, and the Kings had two dominant pairings instead of just the one. They returned to the playoffs with ease, and Doughty even ended up winning the Norris Trophy (probably undeservedly, but I digress).

On the other side of things, the New York Rangers severely limited their season last year by their insistence upon playing their top player, Ryan McDonagh, with possession-anchor Dan Girardi. The two posted a 43.32% corsi share in their 600-plus minutes together at 5-on-5 last season, which was the second worst in the league among pairings that played at least 400 minutes together.

What Does this Mean for the Hurricanes?

In my mind, the importance of these concepts is greater for the Hurricanes than it is for most teams. With the team’s glaring weaknesses in net, possession is at a premium for the Hurricanes. There’s probably a cap on the save percentage that the goaltending on this team is capable of, and thus having defensive pairings that can limit both the quantity and the quality of the chances that Cam Ward and Eddie Lack face is probably the most realistic way for this team to limit the amount of goals they give up.

If they can accomplish the possession and suppression numbers that the following pairings have produced in the past, then the Hurricanes’ goals against totals should decrease to the point where the offensive strides made this season could be enough to finally vaunt them into the playoffs. And at the end of the day, isn’t that all we really want out of this season? Let’s take a look at what some past data says that the Hurricanes should use as their three main defensive pairings throughout the rest of the season. All data used in this post comes from Corsica.hockey and its group statistics tool, which is fantastic and very useful.

Top Pairing: Jaccob Slavin - Justin Faulk

The decision between Ron Hainsey and Jaccob Slavin for the spot with Faulk was a tougher one than I anticipated. Whenever I watch the ‘Canes, my eye test tells me that Slavin’s skating ability and elite stick skills make him the better match as a defensive conscience for Faulk than Hainsey.

The underlying numbers back that idea up, but it’s a much closer gap than I would have expected. The main difference between the two comes in quality driving rather than quantity driving. Between the entirety of last year and the beginning of this season, Slavin-Faulk has been good for a 56.57% corsi share and a 52.54% expected goals share. On the other hand, Hainsey-Faulk has posted a 52.78% corsi share and a 50.29% expected goals share.

Both of those are solid numbers, but the numbers when Slavin is paired with Faulk are clearly superior to the numbers when it’s Hainsey on Faulk’s left side. Hainsey has spent considerably more time with Faulk than Slavin has, but for the numbers with Slavin to come down to the neighborhood of the numbers with Hainsey, there would have to be a rather unrealistic amount of regression involved.

With Slavin and Faulk being 22 and 24 years old respectively, the idea of playing them together now makes a lot of sense. The two could very easily end up forming one of the league’s elite top pairings sooner rather than later. The evidence for this is overwhelming in all three zones already, and both players have even more room to grow and expand their individual skill sets.

Second Pairing: Ron Hainsey - Brett Pesce

This is the toughest pairing of the three to project because they haven’t really spent a whole lot of time playing together yet. However, the styles that the two players utilize suggest to me that they would make a safe, dependable pairing that may not provide a lot offensively but would do well in suppressing the other team’s offense.

The data that we do have available is suggestive of this idea. When the two are on the ice together, their xGA/60 is a stifling 1.95. The idea in having these two players on the ice together would be to create something of a more traditional shutdown pairing that would limit other team’s entries into the zone and turn the puck back for the forwards on the ice to get to work on offense.

Both of these players exhibit solid positioning and defensive instincts, and Pesce is good enough with the puck on his stick to execute that kind of an effective transition game. If prospect Haydn Fleury could be groomed to play the type of game that Hainsey plays but with slightly improved transitional skills and offensive instincts, the philosophy that drives the idea of this pairing could be carried on even after Hainsey moves on from the organization.

Bottom Pairing: Noah Hanifin - Ryan Murphy

For all the flak that Murphy gets from this fan base, this pairing is the very definition of a no-brainer when he returns from his injury that has held him out since training camp.

This pairing was quietly dominant last year. Their xGA/60 was just 2.32, which was only bested (among regularly used pairings) by the Liles-Pesce pairing that absolutely excelled in suppression. Their SA/60 (24.69), FA/60 (37.71), and CA/60 (49.38) were all second on the team behind Liles and Pesce as well.

The company that those numbers put them in league-wide is quite impressive as well. That xGA/60 number was slightly better than that of the pairings of Paul Martin and Brent Burns, Oliver Ekman-Larsson and Michael Stone, Mark Giordano and T.J. Brodie, and Ryan McDonagh and Kevin Klein.

Obviously the competition that those two face would differ to varying degrees from those other pairings, but it’s all relative. As a bottom pairing, Hanifin and Murphy would be able to face other team’s third lines, and they’d likely be able to dominate those match-ups, spending most of their time on the ice in the offensive zone. At that point, the two could both use their elite skating abilities and dynamite offensive instincts to help the forwards generate goals for the good guys all while limiting the chances that their opposition would be able to generate.

The Final Product:

So it’s important to note here that we’re dealing with a reasonably small sample size with just about all three of these pairings. However, when these three pairings are on the ice, the Hurricanes are vastly outperforming their total team averages and the averages of every other pairing that the team has had out there NOT including these three.

The CF% improves by +4.78% with one of these three pairings on the ice compared to what would be expected out of any other pairing. The team’s scoring chance ratio improves by a whopping +6.48%, while the expected goals ratio improves by what is also a significant amount in +5.32%.

So if the data I’ve compiled here is to be trusted, then head coach Bill Peters would be very wise to roll with these three pairings for an extended period of time when Murphy is able to return to the lineup. If the numbers they would put up would come anywhere close to being as good as they have been in the past, then the ‘Canes would go from being a very good possession team to a downright dominant one.

As I mentioned earlier, the small sample sizes we’re dealing with here do dictate that there is some room for regression for any of these pairings. However, that regression would have to be huge for any of them to even approach the struggles of the average of any other pairing that the ‘Canes have iced over the last year and change. Further, if we assume that there is room for regression in a negative sense, then we must also acknowledge the statistical possibility for positive regression.

Now assuming that there is a cap or ceiling at which any player or group of players can carry play, it would make sense for the probability of negative regression to be slightly higher than the probability of positive regression. But once again, the amount at which any of these pairings would have to see their quantitative and/or qualitative possession measurements drop to even come close to the team average OR the average of any other pairing would have to be so significant that for it to happen would be extremely statistically improbable.