Wired Cola

What follows is not original research. Lots of people have trod this ground before and better, but I enjoy the chance to focus on the Canucks and figure out the current situation. I'm especially greatful to Keith Keller for his very useful NHL stats database. It is in many ways superior to the pretty-good database at NHL.com. I used both to support this little essay. You can see my worksheet online, including all my wacky new statistics and crude explanations thereof.

I originally posted almost all of this to alt.sports.hockey.nhl.vanc-canucks and then decided it was worth reposting here.

[referring to the Canuck's January From Hell] I think the Big Swoon is well-explained by the combination of a missing-then-rusty Luongo, the sudden introduction of a key player who spent his first 10-odd games getting up to NHL speed, and a bit of bad luck.

The Canucks were very very bad over that part of the season. Their GDif* in the losing streak was bad: -1.27 over 11 games from Jan 2-31. Points percentage was a miserable .318. To give you an idea, the worst team in the league last year, the dreadful TB Lightnings, had a season points pct of .433. Their GDif was -0.53.**

I ran some numbers on the 2008-2009 season. In that season, every goal you scored more than your opponents was worth, on average, about 0.42 points. The formula applies backwards to goals your opponents score, too. You can envision this as meaning that, on average, a Canucks goal is worth about 4/10ths of a standings point, and a Canucks goal let in is going to cost them about 4/10ths of a standings point. The perfectly average team has a GDif of 0 and gets about 5.5 points every 10 games. During the losing streak, the Canucks were outscored enough to cost themselves, on average, 5.88 standing points relative to the average team's 6.1. In other words, they should have gained less than a point. In fact, thanks to OTLs, they scored 7. OTL points benefit bad teams more than good teams, proportionally, a factor I haven't eliminated using math. That means that a better model would probably have forecast more than 0.2 standings points for the Canucks. 7 is still a stretch.***

Conversely, their GDif during the first part of their win streak was amazing, and has now regressed to merely very good. Their last-10 GDif is 0.8, which again is below the season-long 0.890 level of the 2007-2008 Detroit Red Wings, a team with a .701 points percentage. The Canucks have a .750 points percentage in the last ten, which is a way of saying the Canucks are overachieving in the short term.

The Canucks are still a very good team. To give you an idea of how good, the second-best GDif last year was .488 (Montreal). Over the 69 games of this season, they are a 0.275 team (that would be 6th last year, and is 8th right now), but in the last 20 games they were a 1.05 team.

1.05 means they are HOT! Like, over that 20 games, they have played better than any team played for all of last season, and they have played better than any team has played for all of this season. The Bruins were even hotter over the first half of the season, when I examined them a few months back, but are currently at a much cooler pace.

Conclusion: the Canucks, when coached by Vigneault and featuring their best goalie, an in-form Sundin, and no injuries, are really really good. Remove their goalie and trade good-Sundin for logy-Sundin, and it all goes to crap.

*Goal Differential per Game. Explaining why I use GDif this way takes a bit of work, but the short version is that GDif closely tracks standings points in the long run, and because there are more goal-events than game-events, GDif gives you a less noisy picture of a team's performance level. Extra-short version: it's performance with some of the bad luck removed.

**Last year's league average points percentage was .555. If you want to think of this in terms of circus-points adjustment, a .500 team in the old 2-point system is now a .555 team, and you can adjust your expectations with that in mind. That translates to about 91 points for the 82-game season, or to put it another way, the average team gets about 9 "extra" points through Circus Time each season.

League-wide average GDif was, ahem, 0. I'm embarrassed to admit I calculated that before realizing that GDif is a natural zero-sum and will always be zero in all seasons. It is even zero-sum at all times during the season, of course.

***It's fascinating to watch how the Canucks eased into and out of their losing streak:

http://wombat.san-francisco.ca.us/perl/nhlsched.cgi?season0=2008&op=sched&team=van

Dec.13-Feb 13 (I use "w" for OT wins): LWWLWLLWLWOOLLLLLOOLOWWWWL

Look at how the eases into and out of the 5-game pointless streak on a cushion of OTLs. I think a table of GDif/game would be even more telling, especially if you didn't count OT goals as goals. Note also that the Canucks had no OT wins in that two-month period.