Over at Dustin Byfuglien's entry in the Pilot's Logbook series, the hot topic is Byfuglien the defenseman versus Byfuglien the forward. The argument essentially boils down to a rehash of the age old debate of shot quality and the big mistake. I started to pull some numbers to test my own stance and decided the extent of the information is enough to justify an article of its own.
If you are new to hockey's underlying statistics, there is something you should know: no one has ever thought Corsi is everything. No one.
However, Corsi is very important in evaluating players. Why? Because the object of the game is to score more goals than your opponent. If you do, you win. Corsi tells you who is more likely to outscore their opponent than not. Tyler Dellow once said: it is not that we care more about shots than goals, but that we care more about future goals than past. In other words, we care more about improving a team's chances to outscore, than simply looking at who outscored in the past.
But, is Big Buff an exception to Corsi?
| 2010-14 | Byfuglien | Team | Byfuglien rel to Team | ||||
| Position | TOI | Corsi% | Goal% | Corsi% | Goal% | Corsi% | Goal% |
| Defenseman | 4415.4 | 52.9% | 48.2% | 49.7% | 47.8% | +3.2% | +0.4% |
| Forward | 470.9 | 44.1% | 47.0% | 50.1% | 50.8% | -6.0% | -3.8% |
All numbers are for 5v5 from extraskater.com and stats.hockeyanalysis.com. The values relative to team are just a difference between Byfuglien's values and the team's values during the same stretch of games. This is not like relCorsi% which is the difference with the player on the ice and the player off the ice.
While Byfuglien has been a minus player since he was traded to the Atlanta Thrashers and moved to defense, the franchise has actually been outscored more severely than the big defenseman; a fact often ignored when quoting Byfuglien's plus/minus.
As a fan of how graphical data improves understanding of numbers (believe it or not), here is the same information in visual form:
The y-axis for Corsi% and Goal% is given to show the typical 40-60 sustainable spread seen for NHL calibre players (and teams) where a 50% Corsi rating would be at 0%. The y-axis for relCorsi% and relGoal% is given as Buff – team.
The truth is, Byfuglien on defense improved the Thrashers/Jets goal differential on the long run, despite the fact that Byfuglien (and Enstrom) played the most against players who tend to outscore the opposition. Goal% is highly volatile to outliers and can take a long while to regress.
Corsi isn't everything, but it still is a lot. If you want a simple way to look at it…think of it as a +/- spell-check.
Aside: The improvement in the team’s overall Corsi% coming from moving Byfuglien is highly improbable. Only Chris Thorburn posted a worse Corsi% than Byfuglien as a forward. The improvement most likely extends from a coaching change that occurred one game after Byfuglien was moved to forward. The improvement in Goal% most likely being more minutes given to Al Montoya and Michael Hutchinson over Ondrej Pavelec than Byfuglien being moved to forward. This is why we see the same affects occurring with both Byfuglien on and off the ice after the position change.
Some extra stuff
I had some more stuff on stability of Corsi and regression of goal%, but I've decided to leave that for another time. Here though is one snippet as a preview.
Dustin Byfuglien 2010-14 shot metrics with different defensive partners
| Players | TOI | Corsi% | Goal% | +/- |
| Tobias Enstrom | 2395.3 | 53.6% | 51.7% | 7 |
| Grant Clitsome | 644.4 | 52.6% | 48.6% | -2 |
| Mark Stuart | 410.1 | 47.7% | 31.8% | -16 |
| Ron Hainsey | 394.8 | 51.5% | 52.4% | 2 |
| Johnny Oduya | 387.2 | 53.7% | 45.8% | -2 |
| Zach Bogosian | 245.3 | 49.6% | 50.0% | 0 |
| Keaton Ellerby | 200.5 | 44.3% | 50.0% | 0 |
| Adam Pardy | 198.8 | 53.1% | 33.3% | -5 |
All values are for 5v5 and from stats.hockeyanalysis.com.
Pretty interesting if you ask me.
A side note: a good portion of Ellerby's and Pardy's TOI is when Byfuglien was a forward.
