There are three common themes among hockey statistical analysts. In the beginning, we focus our attention on fixing the plus/minus statistic. After adjusting plus/minus for an assortment of variables, we try to devise an effective defensive measurement. Finally, we attempt to summarize all of a player's contributions into a single statistic. Thanks to Alan Ryder and his Player Contributions rating, in addition to Tom Awad and GVT, I personally never bothered with that third phase.
For those of you not familiar with Ryder and his work at Hockey Analytics, he is one of the top names in this field. On his site you'll find several of his articles, each piece very well researched and in-depth. In 2003, he introduced the first serious attempt at measuring a player's total contribution (offense, defense, etc.) with a single statistic. Player Contribution, or PC for short, is very similar to GVT in both methodology and results, but unlike GVT, PC uses all the currently available statistics, which generally makes it more accurate. The trade-off is that you can't apply PC to different leagues and eras where those statistics are unavailable, so it's meant to be quite specific to modern-day NHL.
Expect the offensive component of PC, known as PCO, to yield results that are very similar to Offensive GVT. As a very general frame of reference, an average PCO is about 8 times an average Offensive GVT. GVT tends to favor puck-moving defensemen and players with reduced ice-time, whereas PC will favor the shootout artists, and the grittier penalty-drawing players, as shown below.
PCO#: Rank under PCO, with the shoot-out component removed
GVT#: Rank under Offensive GVT
Player PCO# GVT# Diff
Dustin Brown 33 193 160
Cal Clutterbuck 242 400 158
Mike Fisher 255 413 158
Nick Foligno 149 281 132
Jarkko Ruutu 245 363 118
Sami Salo 298 199 99
Antti Miettinen 312 210 101
T.J. Hensick 437 327 110
Filip Kuba 262 135 127
Patrice Brisebois 364 232 132
(Minimum 60 games; about 450 such players)
PCO favors checkers and penalty drawers like Brown, Clutterbuck, Foligno and Ruutu, who are valued highly in the PC formula. Alan Ryder explains that "The range of PC adjustments for penalty drawing is generally +/- 10 points. Dustin Brown is the outlier at +20 points. He drew about 20% of all penalties drawn by LA last season. Clutterbuck was +10." Puck-moving defensemen and those that miss a few games, like Salo, Brisebois and Kuba, may get a boost in the GVT approach.
The big differences between PC and GVT are in the defensive zone. PCD, the defensive component of PC, incorporates penalties to a greater degree than does GVT. Defensive GVT appears to give players on the top scoring lines a relative nudge, in the sense that a good offense is the best defense.
PCD#: Rank under PCD
GVT#: Rank under Defensive GVT
Player PCD# GVT# Diff
Ron Hainsey 21 442 421
Bryce Salvador 89 288 334
Andrej Sekera 92 294 329
Brad Stuart 93 345 309
Kevin Bieksa 70 361 291
Evgeni Malkin 381 84 297
Corey Perry 438 133 305
David Backes 441 127 314
Alexander Semin 370 43 327
Blake Wheeler 378 46 332
(Minimum 60 games; 450 such players)
Defensemen who are stay-at-home, or lightly penalized, are cast in a better light in PC, whereas offense-generating forwards like Malkin and Semin look great under GVT's microscope. An average PCD is about 9 times an average Defensive GVT, meaning that there is only a slight preference in how PC will weight a player's defensive contributions overall.
To explore the differences in how PC and GVT measure players, consider Mike Mottau of the New Jersey Devils. In 2007-08, his defensive GVT was 0.6, which is pretty weak, but his PCD score of 44.4 was quite good. In 2008-09, his 5-on-5 goal against average improved from 2.14 to 1.68. While Mottau's defensive GVT shot up that year to a very strong 6.7, his PCD score actually dropped to 36.9. What happened?
My first instinct is to look at penalties, because I know PCD takes a very grim view of those who put their team at a disadvantage, but Mottau's discipline was roughly the same both seasons. The explanation? Penalty killing. Mike Mottau was used more regularly killing penalties in 2007-08, and had a sparkling 5.02 goals against average when shorthanded. Compare that with last season, when he was used less frequently and opponents scored at a 9.18 clip. Special teams play a larger role in PC's assessment than GVT, which is why PCD presented Mottau's 2008-09 season as a step back.
Recently, Gabriel Desjardins wrote an article questioning Colorado's decision to use Kyle Quincey in a penalty-killing or defensive role. Let's take a look at coach Joe Sacco's choices.
RPM: Relative Plus/Minus
ESGAA: Even-strength goals-against average
SHGAA: Short-handed goals-against average
HIT: Hits per 60 minutes
BkS: Blocked shots per 60 minutes
TkA: Takeaways per 60 minutes
GVT: Defensive GVT per 60 minutes
PCD: Defensive PC per 60 minutes
Defenseman RPM ESGAA SHGAA HIT BkS TkA GVT PCD
Kyle Quincey +7 2.77 6.96 4.6 3.0 0.7 0.18 0.88
Adam Foote -7 2.78 7.09 3.9 5.6 1.2 0.02 1.40
Scott Hannan -9 2.79 7.30 1.9 5.4 0.9 0.08 1.94
Ruslan Salei +11 2.86 5.85 4.3 5.8 0.7 0.22 0.93
Brett Clark -4 2.90 10.45 2.6 8.4 0.9 0.11 1.27
John-Michael Liles 0 3.02 9.11 2.2 3.9 0.8 0.10 1.42
Tom Preissing -4 3.68 5.33 2.4 3.4 0.7 -0.01 0.95
According to GVT, Kyle Quincey is the 2nd best defensive player in the bunch, to only Ruslan Salei. It seems quite reasonable to rank him so highly given how many hits he throws, and his low goals-against averages (granted these were for Los Angeles, not Colorado).
However, PCD paints Colorado's blue line in a different color, ranking Quincey (and Salei) dead last among defenders, preferring instead the likes of Scott Hannan. Why? Partly because of penalties. Quincey earned 1.2 minor penalties per 60 minutes and Salei 1.5, compared with 0.4 for Hannan. Additionally, it is penalty-killing time, of which Hannan got plenty: more than double Salei's and triple Quincey's.
One of Gabriel's primary criticisms of Quincey's defensive abilities is that he was sheltered in Los Angeles, being used only against third or fourth lines. Hannan, by contrast, is someone you often see on the ice when Colorado's opponents have their top lines out. Gabriel's own Quality of Competition statistic attests to that: Hannan was 0.07 while both Salei and Quincey were in the negative. While Ryder can't be as open with his precise calculation methods as he once was, I have to believe that quality of opponents still factor into his calculations, helping put a player like Hannan ahead of someone like Quincey.
If you look at all of the available statistics, every NHL player seems to have a certain pattern. In my experience, there are only a couple dozen patterns at most, and evaluating a skater statistically involves finding out which pattern applies to them, and then determining to which level they play that role.
Whenever I'm staring at someone's numbers, and having trouble figuring out the pattern, I find Alan Ryder's Player Contributions invaluable to clarifying the situation. PC helped us understand the changes in Mike Mottau's role, and explained the differences between Kyle Quincey and Scott Hannan where other statistics left the matter unclear. Understanding PC is a key part in developing a detailed, in-depth understanding of NHL players and should be part of everyone's statistical arsenal.
Robert Vollman is an author of Hockey Prospectus.
You can contact Robert by clicking here or click here to see Robert's other articles.