Since the end of the season, a lot of talk has centred around Alex Galchenyuk's step back this campaign. In a year marred by injuries and inconsistency, the 23-year-old forward has been both a lightning rod of criticism and a beacon of hope.
I first wanted to look at the stages of Galchenyuk's development and how Michel Therrien's particular coaching hallmarks had impacted Galchenyuk's performance relative to others in the American's age cohort.
As I delved into the numbers, it became clear that Therrien's foibles — his propensity for line combination switching in particular — affected the entire Canadiens team and not just Galchenyuk. While it's easy to remember Galchenyuk moving up and down the lineup as David Desharnais remained rooted to Max Pacioretty's side, the reality was that Desharnais was as much a victim of Therrien's whims as anyone on the roster.
I've looked at the line combinations utilized by Montreal in the 2014-15, 2015-16, and 2016-17 season under Michel Therrien. Data from after the hiring of Claude Julien was not used for two reasons: 1) the sample size would be too small for independent analysis, and 2) Julien had a legitimate need to juggle his lines in order to better acquaint himself with his new roster.
A basic time-on-ice study would shed light on the number of players that a particular individual had played with over a season, but wouldn't provide insight into whether the deployment was scattered or over consecutive games. Instead, I examined Therrien's player usage both from the perspective of the number of linemates and game-to-game consistency.
I looked at three players: Alex Galchenyuk, David Desharnais, and Tomas Plekanec. Given that much of the controversy behind Therrien's player usage focused on the 1C and 2C positions, looking at the three major candidates for those two positions seemed appropriate.
A "line" was determined by identifying the two forwards who had the largest amount of ice-time with the centre candidate in a particular game. A "combination" was defined as a unique grouping of three forwards irrespective of position; for example, Galchenyuk-Lars Eller-Brendan Gallagher is a different combination from Pacioretty-Plekanec-Gallagher, but not from Eller-Galchenyuk-Gallagher. Finally, a "juggle" was when a line differed from game to game. Switching from combination A to B and back to A counted as two juggles.
Equal opportunity blending
Over the near three-season timeframe, all three Canadiens forwards played on between 10 and 16 different line combinations per season. The annual number of juggles ranged between 17 and 28, averaging out at roughly one line alteration every 3.22 games over the three years.
Unsurprisingly, Therrien's machinations increased as the Canadiens struggled. Both Galchenyuk and Desharnais experienced significantly more roster movement in the 2015-16 and 2016-17 seasons, to the point where Desharnais saw his linemates changed 19 times in 29 games.
Looking at how long these combinations stuck, I grouped line combinations into "stable" and "unstable" categories. Stable line combinations played for four or more consecutive games, whereas unstable ones did not.
Over the three seasons, those three forwards in question played between 36.5% and 45.4% of their games on unstable line combinations. Surprisingly, it was Desharnais who experienced the most turbulence. The data also illustrates a disjointed and frenetic 2016-17 campaign for both Galchenyuk and Desharnais, both of whom played more games in unstable combinations than stable ones.
How does this compare to elsewhere in the league? Well, time and labour constraints admittedly prevent a more comprehensive analysis, but I have brought in two very relevant examples given the current coach of the Montreal Canadiens: 2015-16 data for Patrice Bergeron and David Krejci.
Bergeron played in 11 line combinations, saw his linemates juggled 20 times (once every four games), and played on unstable lines in 25.0% of games. Krejci’s deployment was similar, with even less volatility. The numbers clearly present much higher overall line combination stability for the Claude Julien-led Bruins top six.
Now, one can argue that Bergeron and Krejci are established NHL veterans who have earned their permanent lineup positions, and this is true to an extent. But Therrien moved Plekanec around just as much as Galchenyuk, indicating that seniority was not a significant factor in Montreal’s lineup stability.
Julien was also not afraid to experiment with his top-six wingers. Bergeron played with, in addition to his regular partner Brad Marchand, the likes of Brett Connolly, Jimmy Hayes, David Pastrnak, Lee Stempniak, and Frank Vatrano. Krejci, apart from his regular linemate Loui Eriksson, played with Pastrnak, Ryan Spooner, Hayes, and Matt Beleskey. The difference was that very few of these combinations were one-offs, and a sizable number lasted for multiple games.
An unpleasant smoothie
By now, you're likely asking "what's the point, we know Therrien juggles his lines." The point is that we've never properly examined just what effect that line juggling had on individual player offensive production, and by extension, on team offence as a whole.
Just as a caveat, point production here isn’t tethered to a specific line or even strength situations. The intention of this exercise was not to examine the proficiency of individual line combinations, but to acquire some measure of how constant line juggling affects a player’s overall game at all situations.
When playing on stable lines, Galchenyuk, Desharnais, and Plekanec all showed more production than when playing on unstable lines. This was expected. What was unexpected was the margin: the decline in total production per game ranged from 0.10 to 0.29 points, or a 16% to 45% drop.
Recall that Therrien’s juggling created unstable lines for roughly 40% of all games played, which was twice that of Boston’s top two centres. If we assume that the three players would have continued to produce at the same rate, this time spent on unstable lines, at maximum, cost Therrien an extra 21 points from Galchenyuk (16.3% of total point production), 22 points from Desharnais (17.0%), and nine points from Plekanec (7%) over the course of three years.
Certainly, there is player-to-player variation that prevents making a definitive statement regarding how the blender affected the rest of the forward corps. We can see that the savvy veteran Plekanec is more resilient against line-juggling than the raw youngster Galchenyuk or the one-dimensional Desharnais.
That said, if one assumes that Plekanec is the best-case and Desharnais is the worst-case scenario, we are looking at an overall decline in all-situations forward point production between 7% and 17%.
Montreal’s offence may not be as anemic as initially believed
The struggles of the Canadiens to score goals pre-date the ascension of Michel Therrien and Marc Bergevin. That said, it's evident that the coaching style and decisions of Therrien did no favours in terms of coaxing more goals from his lineup. In fact, the more Therrien swapped his lines around in panic as he searched for wins, the more he dug his own grave. More patience, stemming from a firm understanding of the cyclical and inconsistent nature of goal-scoring, may have resulted in a different outcome in both 2015-16 and 2016-17.
But we do not come here to further bury Therrien. Instead, this data shows that we should view player development during the Therrien years, especially when framed using point production, with a layer of understanding. Certainly, Galchenyuk struggled this year, but he also saw his linemates altered 17 times in 38 games. It is not easy to recover one's game while switching between complementing the styles of Sven Andrighetto-David Desharnais and of Max Pacioretty-Alexander Radulov on an almost nightly basis.
Finally, this information also indicates that the Canadiens are capable of more than what they displayed over the past three seasons, and it offers renewed hope that the Habs can significantly increase their offence just by keeping the blender in the "off" position.
(Stick-tap to Hockeyviz.com for line pairing and point production data.)