Charts & Graphs - Nov. 20, 2013
Keep Calm and Corsi On: outside the playoff picture and looking in
Vancouver assistant general manager Laurence Gilman appeared on the 2nd intermission panel of the Sportsnet broadcast of last night's Panthers-Canucks game, and was asked directly about modern hockey analytics..
From Dan Murphy:
"Fancystats guys say you're doing the right thing. Your Corsi and Fenwick in terms of shot attempts. Do you guys pay attention to that, as a management group?"
For the uninitiated, "fancystats" has become the adopted catch-all term for the Corsi and Fenwick statistics brought up by Murphy. The general rule is that in a small sample, you can better predict what will happen for a team going forward by looking at the shot clock rather than team records. It sounds a little backwards, but let's just say that future scoring chances can be predicted by time of possession, and time of possession can be approximated by the number of attempted shots on goal that a team takes compared to its opposition.
The good news for Vancouver is that after that game against the Panthers, where they somehow managed to get out-shot 36-29 at home, the Canucks are still seventh in the NHL at puck-possession with the score close, having taken 53.5% of those shot attempts. The bad news is that because of the strong Western Conference, the Canucks are sixth in the Conference! Despite the West having just 14 teams, it is becoming more apparent that it will be the more difficult conference to secure a playoff spot.
With three quarters of the season left, which teams in playoff position are vulnerable? Well, both Colorado and Phoenix are in a playoff spot despite being below 50% Corsi Close teams. According to ExtraSkater.com, those two teams are 1st and 6th in the league in "PDO" which is the addition of shooting and save percentage at 5-on-5. It's basically a good measurement of "luck". The Avalanche are scoring on 10.1% of their shots, and the Coyotes at 9.4%, when the NHL average tends to be closer to 8.0%, and there's little indication that teams can repeat high shot percentage numbers over a long stretch of games.
The situation isn't as bad as it looks, even with a five-game losing streak, the Canucks are still doing a lot of the right things. Not against Florida, but the Canucks had out-shot their opponents in the three games leading up to the contest with the Panthers and probably deserved better fates against the Ducks, Sharks and Stars. Any player will tell you that you can't control how the puck bounces, and the Canucks struggle for offence recently hasn't been because of a lack of chances, or shots.
Gilman's answer to Murphy's question, however? Not too revealing. He mentioned that the Canucks research a variety of metrics, which is something we knew anyway.
How does regression to the mean work? Dan Hamhuis demonstrates
At the start of a season there's usually a maligned player that, due to unlucky on-ice shooting and save percentages more than anything, winds up a minus-whatever in the first couple of weeks. This season, it was Dan Hamhuis. If you'll recall, Hamhuis was a minus-4 through six games, capped off with that disastrous own goal against Montreal.
Since then? Hamhuis shot attempt differential rates (Corsi) have continued to be excellent, and after being on the ice for eight 5-on-5 goals against in the first six games, he's been on the ice for just nine in the 17 games since. In fact, his on-ice goals for rate (if you're reading the chart, .50 would be even) is now slightly higher than his Corsi, as it didn't take long for Hamhuis' PDO to regress back to normal:
Through six games, Hamhuis sports a PDO of 885, the worst of any Canuck outside of Alex Burrows, who’s only played one game. (Also sporting low PDOs, by the way: Ryan Kesler and Chris Higgins, both of whom are beginning to get it from the fans for their lack of goal production.) The bulk of Hamhuis’s ugly PDO is the team’s .810 save percentage when he’s is on the ice, the eighth lowest on-ice save percentage in the league among players with at least 5 games. Four of the players with worse on-ice save percentages are on the Edmonton Oilers — that’s how bad it is for Hamhuis right now. This doesn’t mean that Hamhuis isn’t making mistakes in the defensive zone, it’s just that those mistakes are leading to goals against far more frequently than they normally do.
That was written on October 15th.
As of November 20th, Hamhuis' PDO is up to 1001, per ExtraSkater.com. Instead of Canuck goaltenders putting up an .810 save percentage behind him, Roberto Luongo and Eddie Lack have helped out their pal Hamhuis, posting a .900 save percentage now. Meanwhile, the other two players that had been questioned for their lack of production early in the season, Ryan Kesler and Christopher Higgins, appear to be doing fine.
Judging players by starts is a little unfair. If a player goes into a cold streak in January, well, at least you have about 35-40 other games or so to judge him with. A minus-4 slump, or a 5-game goalless drought, looks much worse when it comes in the first six games in the season than it does after a few months of relatively normal production. If it happens midseason, it gets lost in the wash. At the start of the season? It becomes a storyline.
It seems that people have forgotten the slow starts of Hamhuis, Kesler and Higgins. Usually, the best course of action is just to let the percentages work themselves out.
He shoots he scores...efficiently
In the early days, analytics mostly centred around goal and points rates, because there wasn't an awful lot of shots data to parse through. Rather than look at goals or points per game, it made somewhat more sense to look at goals and points per 60 minutes of 5-on-5 ice time, since that took away the advantage from players that played a boatload of minutes.
Here's a quick quiz: Given that Henrik Sedin and Daniel Sedin are first and second on the Canucks in points per 60 minutes this season, who is third?
Second question: Which player leads the Canucks in goals per 60 minutes?
The answer to the first question is Brad Richardson, who has found somewhat of an offensive game in Vancouver (although the Canucks are scoring on 11.7% of their shots with Richardson on the ice, so don't be adding him to your pool team just yet). Richardson has 1.94 points per 60 minutes this season, while he averaged about 1.28 between 2007 and 2013.
The answer to the second question? Zack Kassian is first, then Ryan Kesler, and then Richardson again. The Sedins have been playing so many minutes that even though Daniel is second on the team in total goals with seven, only five have been at even strength, and he plays more even strength minutes than anybody in the league other than his brother. That's going to bring down his rate slightly, but who is to complain as long as he can keep playing over 20 minutes a night?
It's super early, so these numbers have little predictive value, but they're kind of fun to look at.
Incidentally, Alex Burrows has yet to score this season, which is somewhat puzzling. Of the Canucks best single seasons (minimum 60 games) in goals per 60 minutes since 2007, four of the top five belong to Burrows, in 2010-11, 2009-10 and 2008-09. Burrows spoke a little this past week about the "law of averages" and how he usually shoots around 10 or 15 per cent.
That's not entirely true... in just two of his eight seasons has Burrows finished with a shooting rate in that range, usually finishing on the high side. Between the 2008 and 2012 seasons, he shot at an incredible 15.9%, one of the highest marks in the league, and since then in 57 games, he's been at 7.6%. We probably could have expected a slight decline in his personal shot percentage, but it's fallen much lower than we could reasonably expect. He's still generating close to three shots on goal per game, so I'm still looking at him to break it wide open. The Canucks have had some scoring issues lately, but, man, it just doesn't seem like they ought to be.