Rabid Habs

Dissecting Habs Coaching Using Analytics

Jacques Martin Credit: La Presse

Here we are again. The Montreal Canadiens anxiously await the start of training camp after once again falling short of claiming the club’s 25th Stanley Cup. Amongst Habs fans, familiarity has started to breed contempt. With stellar regular season records, the Canadiens have been unable to get out of the Eastern Conference in the playoffs and the blame consistently has fallen on the shoulders of the head coach.

Michel Therrien feels the heat right now; the very same heat that Guy Carbonneau felt when he began coaching in Montreal nearly 10 years ago.

Jacques Martin also had his shortcomings as bench boss of the most illustrious franchise in hockey, as he led the Canadiens to the worst record in the Eastern Conference before the 2012 lockout.

So how much of the success and failure should ride on a coach? In other words, will Michel Therrien receive credit if he leads the Habs to the Promised Land, or can he only be judged by the untimely demise of his club?

For the answers to these questions, we can look at analytics.

Now before you click out of this page, let me explain something: I used to hate analytics. I hated them. I hated the community that tried to over analyze every player and their imaginary numbers that would be used without context to trump every hockey debate. The good news is that I do not hate analytics anymore. I think the main problem with the analytics community is that they do not put their statistics in forms that can be understood without a master’s degree in advanced calculus. As an American literature major, I have found dealing with analytics gurus to be comparable to dealing with science majors. Sure, it is all crystal clear in their mind, but sometimes I feel as though they are speaking a foreign language.

So with that, I will briefly explain a few analytics terms in ways that I think will be easier to understand.

First, we have Corsi. Corsi, simply put, is a shot attempt plus-minus differential. In other words, whenever a player is on the ice, and your team attempts to direct the puck towards the net, regardless of whether the puck misses the net, is blocked or hits the post, you get a plus.

Likewise, if you are on the ice when the opposing team attempts to put the puck on net, you will get a minus. Corsi can be used as a way of defining team success as well as individual success. A Corsi for percentage, or CF%, of 50 is average. Being above 50 means that your team is attempting to shoot the puck at the opposing net more than the other team while you are on the ice.

Next, PDO measures your team’s shot percentage in addition to your goalie’s save percentage. When a team has this number at 1.00, or expressed as 100, it is ideal. If you have a hot goalie and your team is shooting at an unsustainable rate, your PDO will be higher than 100 and will serve as a warning that the team is operating above expectations and could collapse. Toronto led the league in this category in 2013, which was evident by their early exit in the playoffs.

Finally, ZSO% illustrates a percentage of a player’s offensive zone starts. For example, if I step on to the ice to take my first shift and it is an offensive zone faceoff, my ZSO% would be 100. For players, this illustrates how they are used and is expressed as an average. If it is positive, you are in the offensive zone more than the defensive zone. As a team, it illustrates how much they drive the play and strive to keep the puck in the opposing zone.

A final disclaimer: the following stats have been collected from war-on-ice.com. The stats I express will all be at even strength 5-vs-5 hockey unless stated otherwise.

In my coach analysis, I will go in chronological order, beginning with Guy Carbonneau. The former Habs captain was a slam dunk amongst fans at the time of his signing, and some still believe he is the best Habs skipper of the last decade. Before I give my opinion of him, here are some of the numbers:

To put it lightly, the 2006-07 season did not go well for Carbonneau and the Montreal Canadiens. Montreal had a disgusting even strength goal differential of -37 and had an embarrassing CF% of 48.1, good for 22nd in the league. While the results were less than optimal, the Canadiens did have several statistics that were at least partially uplifting. To start, the team was lights-out at home. In Carbonneau’s first campaign behind the bench, the Habs dominated at the Bell Centre to the tune of 26-12-3. Unfortunately, the Jekyll and Hyde Habs were an abysmal 16-22-3 on the road, which gave them no hope of a playoff spot.

The home record indicates, however, that Carbonneau was able to use his players effectively. Because the home team gets the last change, Carbonneau was able to put the lines out on the ice that he thought would give him the best chance against his counterpart’s lines. Montreal would finish 10th in the conference after French goaltender Cristobal Huet was injured.

The 2007-08 season saw Carbonneau’s group fly up the standings in a complete reversal from the previous season. Montreal finished first in the Northeast Division and in the Eastern Conference, but was the success sustainable?

The Canadiens finished the season with an even strength 5-vs-5 goal differential of +6; the 14th best differential in the league. Not exactly championship caliber even strength play. Additionally, the Habs operated at a PDO of 101.5. This indicates an unsustainably high shooting percentage at 5 vs 5 play, which Montreal did seem to have. The Canadiens were 7th in the league in this category at 8.9%.

So what kept Montreal afloat in the regular season?

The short answer is the success of the powerplay. The Habs finished first in the league in this category with the powerplay clicking at 24.06%. While their even strength 5-vs-5 goal differential was only +6, in all situations, the Canadiens finished with a goal differential of +40.

Leading scorer Alex Kovalev finished the season with 84 total points. Pretty good, right? Unfortunately only 33 of those points came during 5-vs-5 play illustrating the importance of the powerplay to the Canadiens. At this rate, it should be no surprise that the Canadiens would fall short of their Cup quest against the Philadelphia Flyers in the 2008 playoffs. The Flyers had the 3rd best CF% while shorthanded during the regular season, meaning they had possession of the puck often while shorthanded and they did not allow teams to take shots when down a man. In short, the Canadiens were average during 5-on-5 play and their powerplay was silenced by the Flyers.

Carbonneau’s coaching career ended just as it began; with mediocrity. Sixty-six games into the 2008-2009 campaign, Carbonneau was canned and then general manager Bob Gainey took over the bench on an interim basis. Through 66 games, the Canadiens were an analytic nightmare. Their CF% measured at 47.7, which is slightly better than the previous season, but still very low, suggesting that the Habs did not dominate puck possession. Montreal also had a pitiful ZSO% of 47.5, indicative of a lack of offensive zone play. Good teams force faceoffs in the offensive zone and maintain puck possession while in the attacking zone; Montreal struggled with both of these.

Again, let’s take a look at Kovalev. Unsurprisingly, only 29 of his 65 points came during even strength 5-on-5 play. Once again, this is indicative of a team that relies heavily on their powerplay.

Also, much like the previous season, the Canadiens were stellar on home ice (24-10-7) but horrendous on the road (17-20-4). As I stated earlier, it shows that Carbo knew how to use his players and took advantage of his last change on home ice. Unfortunately, only 41 regular season games offer this advantage and if your road woes overcome your success on home ice, you’ll find yourself on the outside of the playoff race.

The Habs’ next coach, Jacques Martin, coached the Le Tricolore from 2009 to 2012. Pierre Gauthier’s rebuild brought about more recent players that also played for Michel Therrien, which allows us the opportunity to track the effectiveness of players under different coaches.

In 2009-10 the Habs put up some pretty poor numbers in the analytics department. Here’s how it breaks down:

CF%: 47.1 (27th)
5vs5GD: -11 (21st)
ZSO%: 47.6 (23rd)
PDO: 100.4 (9th)

The high PDO indicates that Montreal had lady luck on their side as they operated with an unsustainably high shooting percentage and Jaroslav Halak standing on his head. This stat explains how the Habs stayed afloat despite horrible possession numbers. Martin’s coaching style is infamous for being boring and these numbers support that idea. These Canadiens sat back, did not push the play and relied on goaltending and shot blocking.

One player I will focus on from this season is Tomas Plekanec, who posted a poor CF% of 46.53 yet put up 70 points. This stat indicates that Plekanec was deployed in mostly defensive situations but was still able to add to the offense. Rochester, New York native Brian Gionta led the Habs in CF% by registering 50.55, which is slightly above average. On another note, PK Subban appeared in two regular season games and put up a CF% of 70.83. Impressive, but it is such a small sample size that it should be taken with a grain of salt the size of Subban’s personality.

The next season saw possession numbers improve for the Habs. Here are the numbers for 2010-11:

CF%: 51.6% (10th)
5v5GD: +1 (T13th)
PDO: 99.6 (19th)
ZSO%: 51.3 (11th)

Montreal appeared to apply pressure and possess the puck more than their opponents, however their even strength goal scoring is still an issue at this time. The highest CF% amongst Canadiens to play at least 60 games belonged once again to Brian Gionta (55.3). Max Pacioretty, who only played 37 games during the 2010-11 campaign before being broken by Zdeno Chara in March of 2011, posted an incredible 59.53 CF% as a 21 year-old. Pacioretty’s efforts in his shortened season, while a small sample size, indicates a player who can play both sides of the puck. The Canadiens would be defeated by the Boston Bruins in double overtime of game seven in round one, despite better possession numbers than this exit indicates and a single goal away from knocking out the eventual Stanley Cup champions.

And then it fell apart. Habs fans everywhere realize that the only good thing to come out of the 2011-12 season was the third overall pick in the upcoming draft that nabbed Alex Galchenyuk. Martin only lasted 33 games before being canned and the Habs numbers seemed to decrease when he left and interim coach Randy Cunneyworth took over. Montreal saw significant decreases in CF%, ZSO%, and 5v5GD after December 17th, which indicates how ill-equipped Cunneyworth was to control an NHL bench. So this season is practically a wash from an analytics stand point. To paraphrase the infamous Dr. Zoidberg, “Your play is bad and you should feel bad!”

Pacioretty had the best CF% on the team with 51.59 while Swiss defender Yannick Weber posted the worst CF% of 45.26 amongst players to appear in at least 60 games. Also interesting to note that while Tomas Plekanec played his way to a 46.49 CF%, he was buried in the defensive zone. On average, the wily Czech vet had 9.69 more defensive zone starts than offensive zone starts per game, which seems to have been the case throughout his career. It is important to note that this was David Desharnais’ first full NHL campaign and he registered a CF% of 48.82, which is indicative of a diminutive forward in his first NHL season. Desharnais’ average was higher than that of his team.

Now the main course: the tanning aficionado and your current coach of the Montreal Canadiens, Michel Therrien. Keep in mind, the 2012-13 season was shortened due to the lockout. Here are the numbers:

CF%: 52.9 (7th)
ZSO%: 51.1 (12th)
5v5GD: +21 (4th)

This season is by far the most analytically successful season the Canadiens have had over the last three coaches. Montreal’s Corsi indicates that they had greater puck possession than their opponents, while the zone starts illustrates that the Habs pushed the play and spent a good portion of time in the offensive zone. Not to mention a top-five goal differential to top everything off.

In terms of players, Pacioretty led all Canadiens in CF% at a whopping 60.48. Desharnais benefitted the most from this, as he improved significantly in CF% by going up to 57.80. Rookie Brendan Gallagher was outstanding, illustrated by his CF% of 57.69 and even Tomas Plekanec improved to 52.16 in this category. Only six Habs had a CF% below 50, including Rene Bourque, who put together a 49.60 CF%.

So why did the Habs have an early playoff exit at the hands of the Ottawa Senators? One simple answer is that Carey Price’s injury was too hard for the Canadiens to make up for in the scoring department. With a PDO of 101.0, the Habs illustrated that their scoring might not be sustainable and the playoffs proved that it wasn’t, as Craig Anderson frustrated the Canadiens to the tune of a five game drubbing.

The 2013-14 season showed a step back as far as possession numbers are concerned, but it wouldn’t stop Montreal from reaching the Eastern Conference final. The Habs decreased significantly in all of the major categories:

CF%: 46.70 (26th)
ZSO%: 44.7 (26th)
PDO: 100.5

Using these numbers, not only should Montreal have been eliminated before the Conference Final—they should have been out of the playoff picture by December. Brendan Gallagher led the team in CF% with a humble 52.05. Subban dipped below 50 in this category (49.86), but this wasn’t a Subban specific issue as only four Canadiens were above 50.00 and one of them, Jarred Tinordi, only played 22 games.

So why did they go so far? A healthy Carey Price keeps the team alive; that goes without saying. Having the 4th best penalty killing unit in the regular season also helps. The long and short of it all is that the Habs overachieved that season.

And finally, the 2014-15 season. Here are the numbers:

CF%: 48.5 (23rd)
ZSO%: 46.9 (25th)
PDO%: 101.6

The PDO stands out to me and it should be obvious to Habs fans why it is so high. Hint: it was not the Habs’ shooting percentage that inflated their PDO. It was Carey Price. When your team’s CF% is lower than lottery-hopeful Arizona, there is a problem. It just means that Michel Therrien’s dump and chase style forces Montreal to play with the puck less and their offense struggles because of it. Gallagher once again leads the charge for the Habs in CF% with a mark of 53.35 with Subban not far behind at 52.09 and Pacioretty at 51.57.

So why does this matter? Well this summer, the Canadiens quietly hired analytic consultant Matt Pfeffer to offer insight to the Canadiens front office. It would be a step in the right direction if the Canadiens were to embrace advanced metrics as there seems to be a direct correlation between teams that have high CF% and teams that have Stanley Cups.

With all things considered, the numbers presented are indicative of coaching systems and it is clear that Montreal is starving for consistent puck possession numbers. Greater possession numbers are usually attributed to teams that give up less goals and have more chances: the last three Stanley Cup Champions have had top five regular season CF%.

So what year was the best? Therrien’s first season was the best by far from an analytic standpoint, but the shortened season might have been too small of a sample size.

My final piece of advice to accompany this stat dump you just sat through is to accept these stats, but not make them the bottom line to describe a team’s success. Sure, it makes sense that if you have the puck more than the other guy, you’re going to win more hockey games. That idea is not wrong. However, hockey is a sport played by human beings that is imperfect by design. You might dominate possession and run into a hot goalie that shuts you out with a final score of 1-0, so a little luck is important too.

Analytics are not going away, and coaches must continue to embrace them to have any chance at winning a Cup. Who knows? Maybe if Michel Therrien recognized that Christian Thomas led the Habs in CF% last season (because he played less than 60 games, I omitted him above), he would have played more than 18 games.

After playing around with analytics to construct this article, I am sure of one thing: I hope Michel Therrien and Marc Bergevin like numbers a whole lot more than I do.

Follow Ian on Twitter: @ihabs1995