Skip to content
Join our Newsletter
VIA store 300x100
Join our Newsletter

Who on the Canucks is driving shot quality?

Some unexpected players come up when you look at expected goals.
Elias Pettersson and Brock Boeser celebrate a goal for the Vancouver Canucks.

Last week, the Vancouver Canucks did something against the Vegas Golden Knights that they’ve struggled to do in recent games: they got scoring chances from in and around the crease.

Three of the Canucks’ four goals were scored from within ten feet of the net, with the only exception being an Elias Pettersson laser beam from the left faceoff circle off the rush.

These were high-percentage chances for the Canucks and, while not all of those types of chances end up in the back of the net, they were more likely than other types of shots. This is intuitive: a shot off a rebound from right on top of the crease is a lot more likely to result in a goal than a wrist shot from the left point.

There’s a difference in quality between the two shots. The shot from right on top of the crease is clearly a higher quality opportunity than the shot from the point. This argument for shot quality is frequently leveled against any analytics that rely on shot quantity, such as corsi, which is simply a measure of shot attempts for and against.

The truth is that shot quantity and shot quality generally correlate quite strongly. Teams that regularly out-shoot their opponent generally out-chance them as well. This can be understood pretty easily from a team’s perspective: teams try to create goals; to create goals, they try to create chances; when you try to create chances, you inevitably take more shots.

In other words, shot attempts are a by-product of scoring chances. If you create more scoring chances than the other team, you tend to out-shoot them as well.

It’s not a hard-and-fast correlation — the ability of some teams to create and prevent scoring chances out-paces their ability to control possession — but it generally holds true on a team level. We can look at this by using an expected goals model, which uses things like shot location, shot type, and whether a shot is off a rebound or the rush, to assign a value to each shot.

Here’s a simplified look at it: an expected goals model looks at a shot and compares it to other shots and says, for example, “20% of wrist shots from 15 feet out in the slot result in goals, so this shot is worth 0.2 goals.”

For instance, Antoine Roussel’s opening goal from last Thursday against Vegas had an expected goals value of 31.62%. That’s the percentage of shots of that type from that location in that situation that resulted in goals. That’s a very high percentage. Of course, the model can’t see that Marc-Andre Fleury wasn’t even in the net at the time of Roussel’s shot or it would have a much higher value.

Essentially, it’s the same shot quantity metric as corsi (or in this case fenwick, looking only at unblocked shot attempts) with a measurement of shot quality to make it a little bit better.

The Canucks this season are 15th in the NHL in score-adjusted corsi percentage at 50.14%. In other words, taking into account the score of the games they’ve been in, they’ve out-attempted their opponents slightly and are near league-average.

In expected goals percentage, the Canucks are 17th in the NHL at 49.78%. They’re a little worse when it comes to creating and preventing chances, but the difference is minimal.

When we look at individual players, however, we can see some significant differences between a player’s corsi and their expected goals. There are players on the Canucks that are significantly better at creating and preventing chances than they are at controlling puck possession, and vice versa.

Let’s take a look at the Canucks that have played in at least 10 games and compare their score-adjusted corsi percentage to their expected goals percentage at 5-on-5.

2019-20 Canucks corsi vs expected goals - dec 23

Where you want to be on this chart is in the top-right quadrant, preferably above the diagonal line. If you’re in the top right, you’re above 50% both corsi and expected goals. If you’re above the diagonal line, you’re providing better shot quality than you are shot quantity.

As might be expected, the Lotto Line is fantastic in both metrics, as is Quinn Hughes. Intriguingly, Miller is a little below the line, suggesting he’s better at driving possession than scoring chances, while Pettersson and Boeser are above the line, but they’re all pretty close: no outrageous outliers here.

The surprise is Loui Eriksson, who pops up in that top-right quadrant. This is from before Saturday’s game, where his numbers took a slight hit, but he’s still up there. Couple notes: this is from a smaller sample size, since he’s been such a frequent healthy scratch, and he’s regularly faced lesser minutes when he’s been in the lineup. Still, there’s an argument to be made that he’s been decently effective when he’s actually been on the ice.

Beyond that cluster of players in the top right, there are some intriguing players that jump out. Chris Tanev and Jay Beagle have the biggest gap between their corsi and expected goals, suggesting they’ve been far more effective than their corsi would indicate.

It would be helpful here to break down their expected goals by for and against, to see where they’ve been most effective.

Player GP xGF/60 xGA/60 xGF%
Elias Pettersson 36 2.63 1.93 57.72
J.T. Miller 36 2.65 1.95 57.66
Loui Eriksson 17 2.44 1.81 57.49
Quinn Hughes 35 2.64 1.99 56.95
Brock Boeser 36 2.51 1.95 56.27
Christopher Tanev 36 2.37 2.05 53.63
Josh Leivo 36 2.51 2.53 49.83
Alexander Edler 27 2.64 2.66 49.83
Tyler Myers 36 2.66 2.71 49.54
Jay Beagle 27 2.44 2.54 49.00
Tanner Pearson 36 2.51 2.64 48.82
Brandon Sutter 22 2.29 2.45 48.37
Bo Horvat 36 2.51 2.74 47.82
Troy Stecher 36 2.16 2.44 46.96
Jake Virtanen 36 2.52 2.87 46.79
Adam Gaudette 26 2.22 2.55 46.58
Tim Schaller 34 2.08 2.53 45.15
Jordie Benn 36 2.12 2.80 43.07
Micheal Ferland 14 1.46 2.63 35.62

Unsurprisingly, the biggest impact Tanev has on shot quality is chances against. His expected goals against is among the team’s best, while actually being slightly below-average on expected goals for.

Beagle, intriguingly, winds up at right around average in both expected goals for and against. Keeping in mind that he is used primarily in the defensive zone — only Tyler Motte among Canucks forwards has started a higher percentage of his shifts in the defensive zone — that’s actually pretty impressive.

Beagle’s corsi percentage is ugly, the worst on the Canucks, but he has actually been shockingly effective at creating and preventing scoring chances.

On the other side, there’s Jordie Benn, whose corsi is bad and his expected goals percentage is worse. With Alex Edler returning to the lineup on Monday, Travis Green is scratching Benn and keeping Oscar Fantenberg in the lineup and it shouldn’t come as a surprise.

Let’s just skip over Micheal Ferland, who fell right off the chart.

Finally, let’s talk about the cluster of players at 50% corsi, but below that in expected goals: Bo Horvat, Jake Virtanen, and Troy Stecher.

Horvat has faced some tough usage and managed to keep decent puck possession numbers, but we can see that he’s struggling to prevent quality chances when he’s on the ice. Horvat thrived in a defensive role last season, but things aren’t going anywhere near as well this season. That’s reflected in his defensive heatmap from HockeyViz.


Bo Horvat defensive heatmap - dec 23 2019



Virtanen has occasionally been used on a matchup line, but has certainly seen easier usage than Horvat, but his struggles defensively are even more significant. He has the worst expected goals against on the team, though his expected goals for is above average. That sounds about right: Virtanen has created quite a bit at even-strength this season, but has given a lot back defensively. He’s strung together some decent games lately, so perhaps he’s turning the corner here, but this is a bit troubling.

Finally, there’s Stecher, whose numbers here could be attributed to teammates more than most. Stecher has played most of the season paired with Benn on the third pairing and you could argue that Benn’s struggles this season have dragged him down.

Stecher’s corsi percentage with Benn is 46.82%; without Benn, it’s 53.59%. The same is true for his expected goals percentage. With Benn, Stecher is at 43.28%; without Benn, he’s at 53.14%.

Ultimately, who is driving shot quality for the Canucks? It’s fair to say that Pettersson, Boeser, and Hughes drive shot quality, as do Tanev and Beagle, albeit in different ways. It will be intriguing to see if they can continue to do so as the season progresses.