Visualizing Corsi and the Canucks

Cam Charron
December 16 2013 01:51PM

In the interest of putting something up on the old blag today and distract from the fact that Canucks fans are yet again in the heat of some silly non-hockey discussion, I thought it might be fun to take a quick look similar to a visual found at mc79hockey's blog that dealt with team Corsi rates when players are on the ice with or without others.

It's a mostly informational thing. Tim Bonnar at Arctic Ice Hockey pointed out that the Winnipeg Jets are a pretty good team when their top players are on, and they're a mostly bad team when their top players are not on. mc79 visualized this by mocking up a graph exploring how the most-used forwards, in order, did with the most-used defencemen. I've made one for the Canucks below.

This chart here is pretty simple. The left-most column is Canucks forwards, sorted in 5-on-5 TOI/60 found over at Extra Skater, and defencemen are sorted along the top the same way from most-used to least-used. The numbers are the Corsi percentages* of the forward with the corresponding defencemen found at Hockey Analysis, and colour-coded red when the Corsi is above 55%, orange when it's better than 50%, or blue when below 50%:

* - [Corsi percentage is the rate of all attempted shots that were taken by the player's team when he's on the ice. If Andrew Alberts is on the ice for 6 shots for and 12 shots against, his Corsi percentage would be calculated as 6 / ( 6 + 12 ) = 33.3%. Corsi percentages have shown to correlate with time on attack .]

Per mc79:

If we assume that the coach uses his best players the most, we would expect that the Corsi% for when the most heavily used defenceman and the most heavily used forward are on the ice to be the highest on the team, barring something really unusual – a defenceman who plays exclusively in the defensive zone or something. … If my theory’s right, things should get worse as we move from more frequently used to less frequently used players.

It looks like this does hold up. The below 50% combinations are generally to the least used forwards on the team. Brad Richardson can be somewhat excused because his role is mostly defensive and thus shouldn't expect to put up very high Corsi numbers. Zack Kassian follows the rule almost perfectly, generating the best Corsi numbers with the most-used defenceman (Bieksa) and the worst with the least-used (Stanton).

The exceptions to the rule appear to be Alex Burrows and Ryan Kesler. Burrows does very well with defensive groupings despite his limited time with the twins this season. (He actually has a higher Corsi when he's away from the Sedins this year, thanks to that excellent run on his line with Mike Santorelli and Chris Higgins). While Kesler gets a lot of minutes, he's not above 55% with any defensive partner.

The Jets had six forwards that had a 50%+ Corsi with the top two defencemen, while the Canucks have seven. The difference between the two looks to be on defence. It's interesting, salary cap-wise, that the Canucks have four medium-to-expensively priced four defencemen, but the cheap bottom-end has been excellent. Chris Tanev was an undrafted free agent signed at the end of the 2009-2010 season. A little over three years later, he's playing on the top defensive pairing with Dan Hamhuis and is still a few days away from turning 24. I think that's a progression nobody could have envisioned, but some injuries in the 2011 season meant he got some looks, and basically earned himself a full-time NHL job with his play that year. He's thrived under John Tortorella, as Thom Drance noted last week.

It would be fun to see more of these, and if they could be automatically generated we could see how other top forward/defencemen combinations do. Bonnar's original theory does seem to hold up though: you'd take Evander Kane and Dustin Byfuglien on the ice and put them up against any forward/defence tandem the Canucks have and they'd handle themselves nicely. The Canucks have much better possession numbers, and win totals, than Winnipeg mostly because a little bit more forward depth and a lot more defensive depth, not necessarily by who is separating themselves at the top of the lineup.

63811cbf517d2d685ea09e103488ea3a
Cam Charron is a BC hockey fan that writes about hockey on many different websites including this one.
Avatar
#1 The Hurt
December 17 2013, 12:30AM
Trash it!
5
trashes
Props
1
props

Why is our world-class city so classless at times? First the Stanley cup street riots. Now some no name attacks Lucic in a bar. Just leave him alone.

Avatar
#2 zipperhead
December 17 2013, 12:35PM
Trash it!
4
trashes
Props
1
props

I am sad because the propeller on my hat is not spinning fast enough, but I am happy because my pocket protector is keeping my pocket nice and dwy

Avatar
#3 Zachary Ellenthal
December 16 2013, 07:12PM
Trash it!
3
trashes
Props
2
props

I threw together the Rangers' chart in the linked blog post:

http://rangersunlimited.com/2013/12/16/visualizing-corsi-and-the-rangers/

Avatar
#4 Peachy
December 16 2013, 03:07PM
Trash it!
2
trashes
Props
10
props

An amusing quote from the Artic article:

"The Jets 3rd line features a player (Matt Halischuk) that is playing for league minimum salary on a 2-way contract. A bottom feeder that plays in the same division as the Jets let him go when he was available for league minimum dollars. They didn’t want him playing on their 4th line and he plays on the Jets 3rd line."

The irony is that Santorelli was given up by the Jets, and he now plays on the Canucks' 2nd line.

Avatar
#5 Lenny
December 16 2013, 03:39PM
Trash it!
2
trashes
Props
1
props

I would love to see how Keith Ballard over the past 3 years stack up.

Avatar
#6 Thomas
December 16 2013, 03:49PM
Trash it!
2
trashes
Props
2
props

@Peachy

You mean now that Santorelli is playing on the Canucks 1st line right?

Avatar
#7 Matt
December 16 2013, 04:39PM
Trash it!
2
trashes
Props
3
props

Also, the Jets have a starting goaltender with a .909 save % compared to the Canucks and their .921+ tandem.

All other things being equal (and they're not), that's half a goal a game right there.

Avatar
#8 Matt
December 16 2013, 04:43PM
Trash it!
2
trashes
Props
2
props

Also:

Holy crap Tom Sestito is bad, and either Dale Weise and Jason Garrison hate playing together or they've just got a really small sample size playing together.

Avatar
#9 Bookie
December 16 2013, 06:32PM
Trash it!
2
trashes
Props
2
props

or maybe both Dale Weise and Jason Garrison hate playing with Tom Sestito?

Avatar
#10 cf
December 16 2013, 08:21PM
Trash it!
1
trashes
Props
0
props

lol

Avatar
#11 Nick Biss
December 17 2013, 03:37AM
Trash it!
1
trashes
Props
0
props

I took the same concept and applied it to the Blue Jackets. The results are a little disturbing when examining the top defensive pairing, but expected since it is Jack Johnson we're talking about.

http://rubberduckhockey.com/2013/12/17/visualizing-corsi-and-the-blue-jackets/

Avatar
#12 Origamirock
December 17 2013, 09:06AM
Trash it!
1
trashes
Props
4
props

While I agree with the idea, I think the execution could use some improvement. I think a gradient (which is made very easy by excel's conditional formatting, for example) is a better way to show the information.

Under the current method, Burrows/Bieksa is the same as Hansen/Stanton. Sestito/Garrison is the same as Booth/Bieksa. Those situations are wildly different.

I think this: http://i.imgur.com/T5ntlyN.jpg

conveys the data much better at a glance.

Avatar
#13 Ruprecht
December 17 2013, 01:13PM
Trash it!
1
trashes
Props
0
props

Neat chart, thanks. The first question that comes to mind is, where is the complementary chart showing how often these players are deployed with one another?

For instance the highest percentage is Stanton and Hank, but I haven't seen them get much ice together relative to our top 4, unless I've missed something. Having a deployment chart to compliment this might help correlate these numbers. It looks impressive, but I find it a little deceiving.

Avatar
#14 JCDavies
December 17 2013, 04:50PM
Trash it!
1
trashes
Props
0
props

@Ruprecht

Agreed. Either there needs to be some way to convey TOI (a second chart, perhaps) or the minimum threshold needs to be increased. 60 min is a really small sample.

Comments are closed for this article.