Western Conference: Line/Tier v. Team Production

Travis Yost
August 01 2013 10:41AM

 

 

INTRODUCTION

One of the elements of hockey analytics that I think is understudied is the midpoint between individual and team-level data. Line combinations change an awful lot over the course of an eighty-two game season. Even in forty-eight, we see coaches -- be it injury-related, performance-related, whatever -- continually juggle trios of forwards on a game-by-game deployment. For this reason, I think a lot of people unfortunately shy away from it.

Rather than exhaust the endless-run of combinations per team and pick through a bunch of data that's short on sample size, I've decided to look at tiers of players based on total TOI. This tends to correlate strongly with the actual, most-regularly used combinations by each coach, but also corrects for a number of potential errors.

One example would be the loss of a player long-term, like, say, Chicago's Patrick Sharp. Sharp, a Quenneville favorite and bona fide top-six forward, missed twenty games during the regular season. As such, he was only available to Chicago for about 58% of regular season games. Another example would be concerning a situation like Ottawa's, where a player like bottom-six staple Zack Smith would log relatively heavy minutes after a run of losses in the depth chart.

Additionally, it tells us just how well teams were able to play with the pieces available over an entire year.

METHODOLOGY

Each team's twelve most-regularly deployed forwards are sorted by even-strength time on ice. Data was culled for each individual at even-strength using Corsi, a proxy for possession-time and scoring chance differential. The number was then adjusted for zone-starts, and then dropped into a weighted tier of most-commonly used forwards. It was then (a) compared to the team's even-strength zone-adjusted Corsi to identify potential outliers; and (b) compared to other team's similar tier.

Not included is the entire Eastern Conference. Comparing the two in a year where zero cross-conference games were played only makes the numbers more noisy.

And, as always, context is important. Especially when looking at data like this.

RESULTS BY TEAM

ANAHEIM DUCKS

 

 

Anaheim kind of suffers from the same thing that ails Pittsburgh over in the Eastern Conference, and that's a lot of top-heaviness. It's interesting to me that, for how much analytics guys killed Anaheim as a team last year, the Winnik-Koivu-Cogliano line was actually pretty damn good.

CALGARY FLAMES

 

 

Pretty standard stuff for Calgary. The team wasn't good, no line was particularly good, and their lowest-crop of talent more or less got killed whenever they were on the ice. Keep dressing Brian McGrattan, though.

CHICAGO BLACKHAWKS

 

 

Chicago's a weird team with these tiers, because guys like Patrick Sharp and Marian Hossa -- as referenced earlier -- kind of slip based on a lack of total ice-time. Either way, Chicago has about nine players that were consistently relied upon last year to crush competition. Even the guys rotating in/out of the lineup managed to get close to 50% of the shot differential battle. This is how you win a Stanley Cup.

COLUMBUS BLUE JACKETS


A lot of people fell in love with Columbus last year as this wild sleeper that sort of came out of nowhere. It's funny how goaltending can play a big role in making a crappy team look not-so-crappy. I've got high hopes for Jarmo Kekalainen, though. Better chance than not this team will head in the right direction.

DALLAS STARS


Fun fact: Reilly Smith's even-strength adjusted Corsi was a rock-solid 54.6%. He's now in Boston. The Seguin/Eriksson swap is going to be talked about an awful lot in the coming years, but there was hardly a word about Smith's inclusion when that deal went down. Maybe there should have been.

DETROIT RED WINGS

 


Another team, like Chicago, that was pretty exceptional at winning the shot battle. They went ahead and added Daniel Alfresdson and Stephen Weiss. Along with Boston, I'd say they're the favorite to win the Atlantic.

EDMONTON OILERS


Good lord. Tambellini's hard-on for team toughness is one of a few reasons why this team absolutely stunk last year. Get away from Taylor Hall et al., and you're pretty much playing an AHL team. Biggest problem for other teams is that Tambellini's gone, and MacTavish seems to know what the hell he's doing.

LOS ANGELES KINGS

 

*picks jaw from floor*

At what point do people start talking about the Anze Kopitar line (along with Dustin Brown and Justin Williams) being one of the best in hockey? Analytics guys have rightfully carried Los Angeles' water for years, but perhaps more time should be doled out to this trio. They're eviscerating everyone. Everyone.

MINNESOTA WILD


This data seems weird to the eye, but Minnesota's problem and search for a second-line is quite real. Away from Mikko Koivu and Zach Parise, only Charlie Coyle really towed excellent possession numbers. A lot of the guys plugged who were shuffled around the middle of the forward ranks (e.g. Dany Heatley, Devin Setoguchi, Cal Clutterbuck) were all sub-fifty.

NASHVILLE PREDATORS


Yeah, this is concerning. Nashville had seven forwards log at least 400:00 TOI at evens last year, and not one came close to breaking even. Patric Hornqvist was well above the mark, but he missed half of the season.

 
PHOENIX COYOTES


It's going to be interesting to see the impact Mike Ribeiro has on this team. The Coyotes, collectively, weren't bad last year. Wonder how the loss of Boyd Gordon impacts them, though. The shutdown center was the team's second-best adjusted Corsi guy last year, trailing only David Moss.

SAN JOSE SHARKS

 


It's like a mini-version of Los Angeles, or Chicago. I think the team had an idea of really shoring-up the bottom-six this off-season for the reasons illustrated above, and Tim Kennedy might remediate that issue to some extent. Someone should really pull Ray Shero's phone cord out of the wall jack, or at the very least, block Doug Wilson's number from incoming calls.

ST. LOUIS BLUES


Lot of balance there. One way to offset having an elite top-line is to have nine, preferably twelve competent guys who can regularly win on margins against their competition. It adds up.

VANCOUVER CANUCKS


I'm going to present that one without commentary. Ryan Kesler can't return to form soon enough. Problem is, I think this team needs like two or three Ryan Keslers.

TIER 1 VS. CONFERENCE

 


Forty-eight games isn't a massive sample, but when you look at the disparity between Los Angeles/Vancouver/Detroit/Chicago and Calgary/Columbus/Nashville -- sheesh.

TIER 4 VS. CONFERENCE


If these bars were buildings, someone took sticks of dynamite, threw it on Edmonton's ground-floor, and watched the structure razed into absolute rubble.

 

73100a182a18dbb0cba9bf8f547ed466
Hockey and hoops. @TravisHeHateMe
Avatar
#1 Kent Wilson
August 01 2013, 12:18PM
Trash it!
1
trashes
Props
1
props

In constructing the tiers Travis, did you use total ice time, or TOI/60 from behind the net, which would be ES minutes/60?

Avatar
#2 Pierce Cunneen
August 01 2013, 12:20PM
Trash it!
1
trashes
Props
0
props

what was the formula you used for zone start adjusted corsi (just curious since they're doesn't seem to be an agreed upon formula out there)?

Avatar
#3 Sean OConnor
August 01 2013, 03:36PM
Trash it!
1
trashes
Props
0
props

Did Colorado get moved to the eastern conference?

Avatar
#4 SmellOfVictory
August 01 2013, 06:02PM
Trash it!
1
trashes
Props
0
props

Interesting. Just a note, the Flames looked a little weird to me so I took a peek at the TOI numbers (both ES and total total), and found them to be misleading in terms of a line composition standpoint (especially because the Iginla trade knocked him into tier 3). Otherwise, an interesting look.

And the overall picture is the same, I realize; it's just that a couple of the bars would be a little bit different if we were talking lines rather than TOI groupings.

Avatar
#5 Elizabeth
August 02 2013, 03:41PM
Trash it!
1
trashes
Props
0
props

Nice, I've never seen Corsi sorted by lines before.

1) Would dividing TOI by games played not remove the Patrick Sharp factor from the tiering? (I think your tiers are actually Toews's line, Shaw's energy line, a Sharp/Bolland/Hossa injury line, and Kruger's line + the healthy scratches

2) Did you have a lower limit of GP or TOI, or would it be worth having 5 tiers? That would show who ices a 4th line of actual NHL players and who is just rotating AHLers and facepunchers between the bench and the press box. Although you would get prospect callups etc in there too I guess.

Avatar
#6 Kent Wilson
August 01 2013, 11:00AM
Trash it!
0
trashes
Props
6
props

Very cool. Nice work Travis.

Avatar
#7 leafnerd
August 01 2013, 12:02PM
Trash it!
0
trashes
Props
1
props

So this data suggests that better teams have better top lines (as measured by corsi) and better depth (as measured by corsi).

Avatar
#8 RexLibris
August 01 2013, 02:22PM
Trash it!
0
trashes
Props
1
props

Amazing work. Awesome. Taking complex and sometimes arcane arguments and laying them out to see in simplicity and honesty is a tough thing to do.

I'd love to see that same thing done at the end of next season to compare. Okay mostly because I'm an Oilers fan and desperately want to see at least a two-storey walkup go up on that tier 4 open-pit mine that is Edmonton in the last graph.

Avatar
#9 Colin
August 01 2013, 04:02PM
Trash it!
0
trashes
Props
1
props

Good thing Brian McGratton is only signed for next year right....... Oh right.

Avatar
#10 OilClog
August 01 2013, 04:45PM
Trash it!
0
trashes
Props
5
props

After seeing this, Kevin Lowe is the smartest man alive. After looking around and coming to the realization that the only way to get this franchise turned around was to STINK! and not just STINK! but that dark dirty deep hole kinda stank. He wondered to himself "how can I do this?" just then Steve Tambellini called needing some advice, two-ply or three-ply?.. Kevin found his answer.

WOW! Steve, please join CBC hockey night in Canada, I think you'll fit right in.

Avatar
#11 Josh
August 01 2013, 04:57PM
Trash it!
0
trashes
Props
0
props

This is really cool. It would be neat to have a webpage to look at these charts and have them automatically update based on the season info so far.

Avatar
#12 Section205
August 02 2013, 04:57PM
Trash it!
0
trashes
Props
1
props

This is very well done and great visual presentation.

As a Flames observer, I am not sure that the team as a whole was as bad (possession-wise) as it looks here.

Behindthenet.ca has the Flames at 9th in West in Fenwick close. Ahead of Ana, Dal, Col and significantly ahead of Nsh, CBJ and lowly Edm.

Duly noted that the Flames absolutely lacked a true top line (#1C obviously) to carry the load. Flames 2nd and 3rd lines seem to hold their own.

Flames also lacked respectable save percentage which sunk them in the goals against and therefore win/loss column.

I would not be surprised to see the Flames be about 9th again (possession-wise) this year. Again without a dominant top line. Unless Ramo turns out to be a decent starting goaltender, the Flames will probably be 13th or 14th in the West standings.

Avatar
#13 Peachy
August 02 2013, 05:17PM
Trash it!
0
trashes
Props
1
props

A comment on the graphs...

Adjusting max and mins for each team graph supports the objective of your comparison, but makes visual analysis more difficult; the visuals become misleading.

The team graph as well is a touch skewed; I'd strongly recommend max and min points to be the same distance from the system's equilibrium of 50%.

Awesome work though. I've never seen anyone do this before and it is enlightening.

Avatar
#14 Peachy
August 02 2013, 05:19PM
Trash it!
0
trashes
Props
1
props

Err, second comment was directed to the conference graph. Oops.

Avatar
#15 Neal S.
August 03 2013, 05:45PM
Trash it!
0
trashes
Props
0
props

@Elizabeth

I've looked at line Corsi information for the Hawks before. TOI isn't going to give you accurate lines for the Hawks. Q double shifts Kane at even strength because he doesn't play on the PK. And Q limited Kruger and Frolik's even strength time for PK duty. But when he used those two at even strength they jumped Shaw's line on the depth chart.

I've found Corsi Rel QoC to be the best criteria to group forwards into their respective lines. Take the top 12 skaters by games played and sort by Corsi Rel QoC. If you do that the information presented here is interesting.

Avatar
#16 Elizabeth
August 04 2013, 10:28AM
Trash it!
0
trashes
Props
0
props

@Neal

Won't QoC put Bolland ahead of Toews though? And any other team that uses a checking line. Yeah, TOI/g isn't fair to Kruger and Fro, but the Hawks being deep enough to have basically 2 3rd lines is not a bad thing.

Avatar
#17 Neal S
August 04 2013, 11:46AM
Trash it!
0
trashes
Props
0
props

@Elizabeth

First off, I look at lines from the defensive side. How teams match up on that end is more important to me than what line numbers the MSM use.

So yes, Bolland's line would be ahead of Toews' line. The Hawks played power vs power and didn't play a checking line this season. They also didn't fight the matchups. So the other team's coaches decided whether Toews or the Kane/Sharp line was the Hawks number one line. They picked Kane.

Now the reason that I use Rel Corsi QoC is because, imo, Corsi needs to be put into the context of who you play against. There is no point in comparing the Datsyuks line's Corsi% to the Crosbys line's Corsi%. When Datsyuk plays against players with positive Corsi his Corsi% is in the 59% range. When Crosby plays against the top players his Corsi% is in the 47% range. Now when Crosby plays against lower lines his Corsi% is in the 64% range. So Crosby's overall Corsi% is higher than Datsyuk's because he plays more often against lesser competition. But as I think I have just shown, that information is misleading.

If however, you show the Corsi% numbers from the defensive perspective, then the numbers, imo, become more relevant. And it also has the side benefit of being more accurate in getting lines that teams actually use. So that would be my recommendation to Travis. Look at this information from the defensive side.

Avatar
#18 Kostamojen
August 05 2013, 12:41PM
Trash it!
0
trashes
Props
1
props

The chart/analysis for the Colorado Avalanche is totally missing from this post. Any chance it will be added?

Comments are closed for this article.