3
Photo Credit: John E. Sokolowski / USA TODAY Sports

Optimizing the Leafs’ healthy defensive pairings for Round 1

So Nikita Zaitsev isn’t playing tonight, thanks to an injury suffered against the Columbus Blue Jackets on Sunday, the last game of the regular season. That’s pretty unfortunate, given how many minutes he’s eaten throughout the year. Because of this, the pairings at morning skate today look just a little different.

Yep. Not a combination of pairings we’ve really seen, outside of some mixed up shifts and some time throughout last year’s blender. If you’re curious, this is how those three pairs have fared over time, relative to their surrounding teammates:

Left Right TOI CF% FF% GF% xGF% SCF%
Gardiner Polak 374.94 -1.35 0.49 7.2 0.33 6.54
Rielly Hunwick 801.63 -3.83 -3.05 -11.5 -2.62 -3.78
Marincin Carrick 144.5 -3.33 -6.18 -5.72 -2.63 2.71

The underlying data looks a little curious when laid out in that way. None of the pairs have really driven play, only one is really expected to come out in the green goals wise, and only one has actually done so. You can at least get the logic behind them stylistically, though; Rielly and Hunwick have familiarity with each other from last year, Polak’s sudden love for forechecking after dump-ins could mesh well offensively with Gardiner’s typical position on the top of the point, and I can’t imagine the third pair plays too much. But out of curiousity, I went through pairs who have played over the past two years, of players still on the team (sorry, Frank Corrado), and tried to figure out the best options under a few different metrics.

To do this, I used Corsica’s combos applet (so individuality isn’t taken into account). All numbers are score-adjusted and team-relative.

Time On Ice

Left Right TOI CF% FF% GF% xGF% SCF%
Gardiner Carrick 803.2 3.54 4.01 10 2.19 3.79
Rielly Hunwick 801.63 -3.83 -3.05 -11.5 -2.62 -3.78
Marincin Polak 394.34 2 0.95 13.94 3.02 3.31

The rationale behind using Time on Ice here is pretty simple; it’s the playoffs, every game matters a lot, and there’s no time to risk on familiarity. This is the most even-strength “chemistry” I could create out of three defensive pairings; players that have played with each other enough over the past two years to know where the other is most of the time.

In the case of the top four, we’re all very familiar with what each pair brings to the table. Gardiner-Carrick has been Toronto’s sheltered fancy stat pair this year with a bit more emphasis on gaining the offensive zone than protecting the defensive one, though they have managed to keep shots away there too. Rielly-Hunwick didn’t work out too well last year, but they played almost all of the “tough” minutes on a 30th place team that was devoid of top end forward talent.

Marincin-Polak looking not awful is a surprising one. Both play awkward-looking hockey and have struggled in similar ways, but for whatever reason, they’ve been Toronto’s best pair at limiting shot attempts and goals in the past two years, and the second-best at limiting expected goals.

Shot Attempts

Left Right TOI CF% FF% GF% xGF% SCF%
Gardiner Marincin 109.64 6.11 4.91 8.31 5.98 6.27
Rielly Carrick 207.9 0.63 0.05 -5.73 6.44 20.48
Hunwick Polak 790.46 -2.95 -1.74 1.69 -0.17 -0.08

Looking at good ol’ fashioned Corsi is a little tricky if you’re trying to make three pairings. Every above-water pair option involves one of Gardiner or Rielly, with the combination of them both leading the way. With that in mind, Hunlak sticks together to make this work; which might not be the worst thing if they continue to play the way they have in the past few weeks, taking more offensive risks and carrying the puck more.

I feel like the crowd that would cheer on that decision the most would also set fire to the ACC the second they saw a Gardiner-Marincin pairing, though, so we’re going to leave this one to the side.

Expected Goals / Scoring Chances

Left Right TOI CF% FF% GF% xGF% SCF%
Gardiner Marchenko 71.84 -3.73 -1.54 35.24 6.61 16.82
Rielly Carrick 207.9 0.63 0.05 -5.73 6.44 20.48
Marincin Polak 394.34 2 0.95 13.94 3.02 3.31

Shot quality isn’t everything, but it’s definitely more than zero, so using Corsica’s expected goals model to find the pairs that ride the probability wave the best is worth doing. Perhaps not surprisingly, this is the closest to optimal I think we’re going to get using the sort by spreadsheet method. I don’t really trust 72 minutes of Gardiner/Marchenko is a sure thing (especially given Marchenko’s near-opposite results with Rielly), and I don’t know if I like the idea of scratching Matt Hunwick during a solid run of form, but going with this arrangement gives you a 3/3 left and right shot balance.

The interesting thing about going down this road is that Washington is a little weaker at turning their attempts into legitimate changes than most teams are. While still a Top 10 team in expected goals (7th) and chances (9th), its a drop off from their 3rd seed in Corsi/Fenwick/Shots on Goal. Toronto swings the opposite way, ranking 10th in Expected Goals this year and 5th in Scoring Chance Percentage. Perhaps making this series about the danger areas could make an impact. I wouldn’t mind trying it. We almost assuredly won’t see it.

Actual Goals

Left Right TOI CF% FF% GF% xGF% SCF%
Gardiner Marchenko 71.84 -3.73 -1.54 35.24 6.61 16.82
Hunwick Marincin 59.44 -4.89 -4.6 16.45 -3.83 -5.43
Rielly Polak 200.73 -5.01 -4.19 10.93 2.79 6.44

I’m by far the least confident in this arrangement, but I’m including it out of solidarity. None of the pairs are really familiar with each other. Some have done okay from an xG/SCF% percentage, but all three have been shelled in terms of giving up attempts. I’d bet that these three pairs are the pairs at hand based on shooting/save percentage luck and a lack of time to turn it around.

Conclusion

Trying to figure out how to arrange your team’s defensive core when you lose your consensus top defenceman on a specific side is very difficult, especially when you’ve avoided having to test out how such a thing would work throughout the season. Ultimately, I don’t hate tonight’s arrangement too much. It could be better, but it seems like Babcock is going for a mix of familiarity, size with skill, and hot-handedness, and if it doesn’t work, they’ll adjust again quickly. I would personally feel most confident trying the xG/Scoring Chance pairs, but you’d be hard pressed to find an arrangement of six guys on most NHL rosters that would have me fearless against the Capitals anyway.