Who should play with Elias Pettersson: Weighing in with analytics

Pass it to Bulis

Even the most optimistic projections for Elias Pettersson’s rookie season expected a few more growing pains for the rookie centre. But even as fans can see Pettersson struggle at times, he just keeps piling up points.

Pettersson is now seven goals and 13 points ahead of the next best rookies in both categories. Heck, he has more points at even-strength alone than the next best rookie has in all situations. Despite missing six games, he’s leading the Canucks in scoring and is tied for 14th in the NHL in goals.

article continues below

It hasn’t mattered too much who he plays with: he’s scored at a first-line rate no matter who lines up on his wings. But which wingers bring out the absolute best in Pettersson? Who should stick with him throughout the rest of the season as he vies for the Calder Trophy as rookie of the year?

That’s not an easy question to answer and it’s one that head coach Travis Green has to balance with the needs of other lines and players. One way to look at the lines, however, is through the lens of analytics. This won’t necessarily be a thorough analytical breakdown; it’s more a piece of the puzzle that can be looked at in conjunction with other statistics and the eye test.

I’m going to be looking primarily at two statistics from Corsica.Hockey: corsi and expected goals.

Corsi is a very basic statistic: it’s simply shot attempts when a player is on the ice. It is mostly used as a percentage at 5-on-5. A percentage higher than 50% indicates that your team has more shot attempts than the opponents when you are on the ice.

Expected Goals is similar, except it assigns each shot attempt a likelihood of becoming a goal based on elements like the type of shot, the distance from the net, the angle to the net, and whether the shot was on a rebound or off the rush.

By using both statistics when looking at each line, we can see whether that line is out-shooting the opposition when they’re on the ice and also whether they are creating and preventing quality scoring chances. Essentially, we’re looking at both shot quantity and shot quality.

Here are all the lines that have played at least 19 minutes together this season. For some of these lines, that means a pretty small sample size, but it’s still interesting to look at how they’re performing in that limited sample.

Roussel Horvat Eriksson 21.50 68.97 81.13 12.16
Goldobin Pettersson Virtanen 39.87 60.00 52.61 -7.39
Roussel Horvat Gagner 30.20 58.82 62.81 3.99
Schaller Beagle Motte 24.30 58.54 81.78 23.24
Leivo Pettersson Boeser 42.22 57.35 71.38 14.03
Goldobin Pettersson Eriksson 94.28 55.98 51.96 -4.02
Eriksson Granlund Virtanen 27.45 55.32 64.44 9.12
Leipsic Horvat Eriksson 31.67 54.55 47.98 -6.57
Schaller Granlund Motte 64.13 52.83 53.71 0.88
Goldobin Pettersson Boeser 78.73 52.41 56.13 3.72
Schaller Horvat Granlund 19.88 51.61 54.00 2.39
Leipsic Gaudette Archibald 24.77 50.00 46.27 -3.73
Roussel Granlund Eriksson 35.55 50.00 32.98 -17.02
Roussel Sutter Virtanen 49.88 47.83 53.87 6.04
Goldobin Horvat Boeser 38.85 46.67 57.03 10.36
Schaller Beagle Granlund 21.33 46.43 65.47 19.04
Schaller Horvat Motte 21.48 45.65 36.67 -8.98
Baertschi Horvat Boeser 52.62 43.02 47.30 4.28
Roussel Granlund Motte 26.83 42.22 60.29 18.07
Roussel Sutter Eriksson 28.68 41.67 34.81 -6.86
Schaller Horvat Virtanen 22.00 40.48 27.54 -12.94
Roussel Gaudette Eriksson 21.55 37.14 33.33 -3.81
Goldobin Horvat Virtanen 35.53 36.11 30.55 -5.56
Granlund Horvat Virtanen 19.10 34.21 28.81 -5.40
Motte Sutter Virtanen 20.68 25.00 16.59 -8.41

Hopefully that’s not too overwhelming. I’ll break it out into just Pettersson’s lines in a moment. First, a few observations.

These statistics are colour-coded: yellow is high, red is low. For time on ice, that just means that the lines highlighted with yellow have played together more, while the red have played less. For CF% and xGF% (corsi-for percentage and expected goals-for percentage), yellow is good and red is bad. For the final column, it just indicates the difference between their CF% and xGF%. More yellow means their shot quality exceeds their shot quantity in some way.

Some interesting lines show up near the top of this list.

The line of Antoine Roussel, Bo Horvat, and Loui Eriksson has dominated the opposition with a ludicrously-high 68.97 CF% and an even higher xGF%. When they’ve been on the ice, the puck has largely been in dangerous areas in the offensive zone. Keep in mind that they’ve played just under 22 minutes together, but that’s a line that should see some more time together.

Surprisingly, the fourth line of Tim Schaller, Jay Beagle, and Tyler Motte is near the top in CF% and has the highest xGF% on the Canucks. We’ll see if that lasts — they had a tough time starting the game against Columbus's first line on Tuesday — but it’s intriguing to see such positive underlying numbers from that trio.

There are some interesting lines at the bottom too. While Jake Virtanen has taken a big step forward this season, it really seems like he shouldn’t be playing with Bo Horvat. The three lines that feature those two forwards are in the bottom five in corsi percentage and are even worse in terms of expected goals. He’s worked much better with other wingers.

There’s also an argument against reuniting Sven Baertschi with Horvat and Brock Boeser when Baertschi returns to the lineup.

But let’s isolate the Pettersson lines and see what the analytics have to say.

Nikolay Goldobin Elias Pettersson Jake Virtanen 39.87 60.00 52.61 -7.39
Josh Leivo Elias Pettersson Brock Boeser 42.22 57.35 71.38 14.03
Nikolay Goldobin Elias Pettersson Loui Eriksson 94.28 55.98 51.96 -4.02
Nikolay Goldobin Elias Pettersson Brock Boeser 78.73 52.41 56.13 3.72

There are a couple intriguing things here. One positive is that all four lines are solidly above 50% in CF%. While Pettersson himself has a CF% just under 50% (49.64), when he’s on the ice with his regular linemates, the puck moves in the right direction.

In terms of expected goals, Pettersson’s newest linemates have been his best. Josh Leivo has been a good fit with Pettersson and Boeser, freeing up the puck on the forecheck and getting to the front of the net to screen the goalie and tip the puck on net.

For contect, among lines league-wide that have played at least 100 minutes together this season, the highest xGF% is 71.49%, and that’s the Pittsburgh Penguins’ top line centred by Sidney Crosby, with Dominik Simon and Jake Guentzel on his wings.

In other words, if the line of Leivo, Pettersson, and Boeser are able to sustain their current xGF%, they would be up there among the best in the league.

Goldobin, Pettersson, and Boeser have the second-highest xGF% of these four lines, but the lowest CF%. That line has been better at the shot quality part of the equation than the shot quantity.

The line with Nikolay Goldobin and Jake Virtanen has the highest CF% of the four lines, but has the second-lowest xGF%. That’s a pretty significant drop and it perhaps speaks to what has frequently been observed about Virtanen: his penchant for low-percentage shots from the outside.

On the other hand, the line with Goldobin and Eriksson also has a lower xGF% than CF%, so maybe that’s not it. So let’s break out the for and against parts of corsi and expected goals to see where each of these four lines is doing well or poorly.

LW C RW TOI CxGF CxGA xGF xGA xGF - CxGF CxGA - xGA Difference
Leivo Pettersson Boeser 42.22 1.59 1.18 2.22 0.89 0.63 0.29 0.92
Goldobin Pettersson Boeser 78.73 3.09 2.81 2.93 2.29 -0.16 0.52 0.36
Goldobin Pettersson Virtanen 39.87 1.83 1.22 1.31 1.18 -0.52 0.04 -0.48
Goldobin Pettersson Eriksson 94.28 4.19 3.30 4.38 4.05 0.19 -0.75 -0.57

I’m comparing the two statistics — corsi and expected goals — directly with a technique I used last season. On average, a shot attempt has a 4.07% chance of becoming goal if you ignore every aspect of shot quality. By multiplying corsi by 4.07%, we get something I call Corsi Expected Goals.

Think of it as a simplified version of expected goals. Whereas xG treats every shot on its individual merits — location, shot type, angle, whether it’s off the rush or a rebound — giving each attempt a unique likelihood of becoming a goal, CxG treats every shot attempt as having the exact same likelihood: 4.07%.

The upshot is we can compare whether a line is creating better quality chances than you would expect from their corsi, or preventing the same defensively.

If that’s confusing, let’s keep it simple: the line of Leivo, Pettersson, and Boeser has created better quality chances than you would expect from their corsi, but have also prevented quality chances defensively, albeit to a lesser extent.

Meanwhile, the line with Virtanen has, in fact, created poorer quality chances than you would expect given their corsi.

The line with Eriksson, on the other hand, has suffered mainly defensively, which is unexpected. They’ve given up better quality chances than you would expect.

So what does this mean? Keep in mind, all four of these lines are good options and Pettersson has scored at a first-line rate with all of these linemates.

The best line so far has been Leivo, Pettersson, and Boeser, but keep in mind they have a smaller sample size together. Replacing Leivo with Goldobin is a solid option as well: they have played very well together in a larger sample size

As for Eriksson, perhaps it would be best if he stuck with Horvat.






Read Related Topics


NOTE: To post a comment you must have an account with at least one of the following services: Disqus, Facebook, Twitter, Google+ You may then login using your account credentials for that service. If you do not already have an account you may register a new profile with Disqus by first clicking the "Post as" button and then the link: "Don't have one? Register a new profile".

The Vancouver Courier welcomes your opinions and comments. We do not allow personal attacks, offensive language or unsubstantiated allegations. We reserve the right to edit comments for length, style, legality and taste and reproduce them in print, electronic or otherwise. For further information, please contact the editor or publisher, or see our Terms and Conditions.

comments powered by Disqus