April 5, 2012; Raleigh, NC, USA; Montreal Canadiens center David Desharnais (51) carries the puck against the Carolina Hurricanes at the PNC center. The Hurricanes defeated the Canadiens 2-1 in the shoot out. Mandatory Credit: James Guillory-US PRESSWIRE
If there is one advantage of an impending lockout, its that there is going to be plenty of time to fully understand what happened last season.
With that in mind I have new ways to break down the recent numbers for the Habs current players, in what I’m calling “Components of Offense.”
Offensive play in hockey is all about goals. The only way a player helps you offensively is if he gets your team more of them, by direct (i.e. scoring or setting up a goal) or indirect (i.e. forcing an offensive zone faceoff which allows your best line to come on and then score). But goals can be a sticky thing to work with in hockey stats. Simply put, there aren’t enough of them for accuracy. The simple behaviour of relatively small numbers (like the 15 or so goals a player might get on even strength per season) in a probabilistic system (playing a game involving a bouncing cylinder on a slippery surface qualifies) is going to be effected by random volatility because relatively good or bad fortune can have a big proportionate effect on outcomes. So hockey stats tend to break goals into its two components, shots on net and percentage of shots that go in the net. The much greater number of shots allows for better data analysis, especially given the findings that shot measurements are fairly constant in hockey while shooting% is heavily effected by chance and circumstance.
Simple right? Well, with all ideas that are both simple and useful, in execution simple becomes a lot more complicated. There are natural differences in players and lines shooting percentage according to talent and style of play that can throw off shots based analysis. Figuring out exactly how to compensate for that while retaining the core power of shots based analysis would be a big step forward for the hockey analytical community.
So with that in mind I have looked at breaking down individual and team on ice shots to better understand what goes into the creating offense. To this end I’ve run the 3 year averages of player personal and team shots per minute rate to work out the relationships that go into aggregate goals scored.
First we should establish what is typical performance for a position. Last year on 5 on 5 play forwards take 73.1% of the shots and 87.3% of the goals, while defensemen take 26.9% of the shots and 12.7% of the goals. For individuals this means forwards take a average of 29.1% of the team’s on ice goals and 24.4% of the shots while a defenseman 13.4% of on ice shots and 6.3 of the goals. Forwards get goals on 9.61% of their shots while defense get goals on 4.01% of shots, with an overall average on ice shooting percentage of 8.17%. From these results we can derive a positional player’s typical team-mate shooting percentage, 7.62% for forwards, 8.74% for defensemen.
Today we use this method to look at the Canadiens centers:
| Center | Personal shots per 60 |
Shooting%
Goals Scored per 60
Teammate Shots per 60
Teamate Shooting%
Team On Ice Shots per 60
Team Goals per 60
Percentage of Team Shots
Plekanec
8.40
9.39
0.79
20.04
7.54
28.44
2.30
29.54
Desharnais
4.20
13.89
0.58
27.03
8.46
31.23
2.87
13.45
Eller
5.76
10.00
0.58
20.94
6.32
26.70
1.90
21.57
Gomez
6.28
4.28
0.27
26.63
6.65
32.91
2.04
19.08
The most dangerous goal scorer of the bunch is obviously Plekanec by a considerable margin, bolstered by a very respectable shot rate. Eller and Desharnais are pretty much on par for goal scoring, Desharnais low shot rate compensated for by a sky high personal shooting percentage. Maintaining a ~14% shooting percentage is relatively rare, but Desharnais is a low volume shooter that passes up shooting opportunities which is similar to two established high % shooters in Alex Tanguay and Sergei Kostitsyn. Shot creation tends to peak by 23, so Eller is the only player here expected to up his rate on talent progression and not by much. He’s played shutdown minutes with generally poor offensive linemates however so we may see some progression due to more favourable circumstances. Gomez’s vanishing ability to get the puck in the net has been the downfall of his goal totals, although he more that 3 years ago his shot taking rate was similar to Plekanec’s.
For team-mate shots we can see a sharp distinction between the playmaking (Gomez and Desharnais) and two-way centermen (Eller and Plekanec). With that high a rate for Gomez and Desharnais it looks like those two really are increasing shooting chances for their linemates. With average on ice shooting, the 6 shots per hour more would be worth about .45 goals per hour. However, keep in mind that Desharnais most common linemates have been Pacioretty and Cole, two very high volume shooters before they played with Desharnais. How much of Desharnais or Gomez’s increased on ice shot production is them and how much is their linemates is a fair question. Gionta’s shot rate per 60 improved by .45 or 5% when coming to Montreal and playing with Gomez the next two seasons. Cole’s shot rate improved by 1.36 per 60 or 17% after coming to Montreal and Pacioretty’s improved by 1.00 or 9.3% in 2011-12 primarily with Desharnais rather than 2010-11 primarily with Gomez. This kind of smaller improvements suggest to me that the playmakers are improving teammate shots but a big component is quality of linemate.
I think its no coincidence that the highest team-mate shots centers played with the very high volume shooters in Cole, Pacioretty and Gionta, while the lowest team-mate shots center in Eller’s top linemate was the offensively inept Travis Moen.
That said, we could also take this to mean that Desharnais is a slightly superior playmaking center to Gomez and Eric Staal. That’s no mean feat. While I don’t think he’s the driver on the bus of last year’s 1st line, providing comparable ability to set up plays to two veteran centers with high career assist totals would be nothing to sneeze at. But if he was a real 1st line center like some of his proponents are claiming it would have to be either attached to some above average native goal scoring talents of his own or be so good as a set up man he’s drowning in assists.
Teammate shooting percentage is a big cause for the spread in on ice offense as well with Desharnais’s 8.46 carrying him to the highest on-ice scoring and Gomez and Eller’s lower value dropping their team offense. Luck and teammate quality are probably the dominant factors here. Most research into team-mate on ice shooting shows that players have very little influence in how well their linemates convert their shots, with even the league’s top players having a relatively small effect. Meanwhile, factors outside a player’s control in randomness and differences in natural shooting talent is large. These results probably reflect how Desharnais was playing with a typical 12.2% even strength shooting% forward in Cole while Eller’s most common winger has been 7.1% Moen.
The proportion of team shots each player takes is also interesting. It looks like the typical value for Habs centers is about 20%, meaning centers are average in shot taking amoung the 5 players on ice. Plekanec buck the trend high and is a major factor in his team’s shot taking while Desharnais takes about as many shot as a defenseman, leaving the shooting duties to his wingers.
Quality of Competition (corsiRel) |
Average Zone Start Ratio
Plekanec
0.755
46.7
Desharnais
-0.242
50.5
Eller
0.200
49.4
Gomez
0.400
55.7
One factor that has been missing from our discusion is difficulty of opposing circumstances, which we look at here by the two standard methods Quality of Competition measured by opposing relCorsi and percentage of offensive to defensive zone faceoffs (OZone Start%)
Its clear that Plekanec consistantly faces the heaviest minutes. He’s the most counted upon in the defensive zone and faces the toughest opponents. This is likely a big reason that his team-mates offensive performace lags. Meanwhile Desharnais faces the lightest minutes in competition and relatively high offensive zone use, reflecting minutes designed for scoring, which is surely a factor in his good team production. Eller has been moving towards Plekanec’s role but doesn’t fill those shoes yet while Gomez once was a tough opposition player but has fallen down the matchup depth chart.
I’ve shown a bunch of rate stats so players can be easier to compare to one another, but what do these numbers mean in real terms? To answer that I have converted the personal and team goal scoring into what would result from this player performing at this level in top six minutes (14 minutes ES per game) over 82 games.
Goals per 82 |
On Ice Goals per 82
Plekanec
15.1
44.0
Desharnais
11.2
54.9
Eller
11.0
36.4
Gomez
5.1
39.0
Plekanec has been a player that will get to 20 goals on the year with special teams time. Likewise Eller and Desharnais have shown they can hit 15 with some goals elsewhere. Gomez has been inept in goal scoring. Desharnais has been a part of strong offensive units, Plekanec has been generally good while Gomez and Eller’s lines have struggled to score.
Thanks to behindthenet.ca and stats.hockey.analysis.com for the databases to make this work possible.