Post-Lockout Divisional Imbalance
by Hawerchuk on Jun 9, 2010 9:00 AM EDT
Email
PrintEver wondered about the extent of divisional imbalance in the NHL? We know that the southeast division isn't very good, but how bad is it? Here's the Corsi% (even-strength shots only) with the score tied every year since the lockout:
| 2005 | 2006 | 2007 | 2008 | 2009 | Total | |
| atl | 50.6 | 47.6 | 50.5 | 50.4 | 51.4 | 50.2 |
| ne | 50.2 | 50.1 | 49.3 | 49.5 | 51.3 | 50.1 |
| se | 48.6 | 50.4 | 49.6 | 48.6 | 47.4 | 48.8 |
| cen | 50.2 | 50.4 | 53.5 | 52.9 | 53.7 | 52.3 |
| nw | 49.3 | 48.8 | 46.6 | 47.7 | 46.0 | 47.6 |
| pac | 51.2 | 52.9 | 50.5 | 50.9 | 49.8 | 51.0 |
The southeast division is indeed bad, but the northwest is even worse. How much of that is having to play more games against the much stronger teams in the central and pacific divisions?
| atl | ne | se | cen | nw | pac | |
| atl | 49.5 | 51.2 | 50.1 | 50.6 | 48.4 | |
| ne | 50.5 | 50.5 | 47.0 | 50.3 | 49.1 | |
| se | 48.8 | 49.5 | 45.7 | 50.9 | 47.1 | |
| cen | 49.9 | 53.0 | 54.3 | 53.4 | 51.1 | |
| nw | 49.4 | 49.7 | 49.1 | 46.6 | 47.4 | |
| pac | 51.6 | 50.9 | 52.9 | 48.9 | 52.6 |
The NW division is slightly better than the SE division against all shared opponents. But SE division teams outshot NW teams in head-to-head games. The difference between the two divisions is negligible, though the NW's stronger showing against the pacific and central suggests that it's just a little bit better than the SE.
60 comments
|
0 recs |
Do you like this story?
Comments
I blame Edmonton
The New Improved Avalanche. Now with Real Coaches!
Jibblescribbits: C'mon over and waste some time
guys, guys…the wild and flames are also pretty bad.
Join me on the Hockey Blog Adventure! (or Twitter.) GO BRUINS! (and Wild!)
by Cornelius Hardenbergh on Jun 9, 2010 11:18 AM EDT up reply actions
Which leaves Vancouver as Washington NorthWest.
Kinda fitting, they are just over the border.
Join me on the Hockey Blog Adventure! (or Twitter.) GO BRUINS! (and Wild!)
by Cornelius Hardenbergh on Jun 9, 2010 11:19 AM EDT up reply actions
Good point
I blame the Wild, and Edmonton
The New Improved Avalanche. Now with Real Coaches!
Jibblescribbits: C'mon over and waste some time
by Jibblescribbits on Jun 9, 2010 5:07 PM EDT up reply actions
I blame Colorado and Ryan Johnson.
Editor of The Copper & Blue, and leader of The Cult Of Hartikainen.
what do you have against Darcy Hordichuk?
by Passive Voice on Jun 9, 2010 10:04 PM EDT up reply actions
I predict that this post will cause an uproar among those NW Division teams who take pride in how tough their division is.
All things Thrashers + stats: www.birdwatchersanonymous.com
I think (hope) most folks realize that the NW division is pretty bad right now. In our NW preview at C+B I had it as the worst division in the Conference and didn’t get any disagreement IIRC.
by Scott Reynolds on Jun 9, 2010 1:19 PM EDT up reply actions
yeah
the NW is pretty bad at the moment. I thought it was pretty good a couple seasons ago, but that doesn’t appear to be the case.
The New Improved Avalanche. Now with Real Coaches!
Jibblescribbits: C'mon over and waste some time
by Jibblescribbits on Jun 9, 2010 5:08 PM EDT up reply actions
It has never been great IMO. The Flames have improved somewhat since the lockout (probably their worst season was the season they won the division) while the Oilers declined, and then Colorado took a big step back with the loss of Sakic and Smyth. And VAN was never as good as its record suggested.
Just a pile of shit that we watch, unfortunately.
by R O on Jun 9, 2010 5:10 PM EDT up reply actions
Just after the lockout the divison was pretty good. In 2005-06, every team had a positive goal differential. 2006-07 was similar until the Oilers completely collapsed. As a group, they’ve regressed significantly and they’ve never had one of the dominant teams, even when they were good. It was just a bunch of teams between 3rd and 10th.
by Scott Reynolds on Jun 9, 2010 10:27 PM EDT up reply actions
And last/this season the NW was pretty awful.
The New Improved Avalanche. Now with Real Coaches!
Jibblescribbits: C'mon over and waste some time
by Jibblescribbits on Jun 10, 2010 9:56 AM EDT up reply actions
Absolutely right! Those middling teams just after the lockout are getting dragged down by some downright bad teams now.
by Scott Reynolds on Jun 10, 2010 11:42 AM EDT up reply actions
It’s sad that the Maple Leafs are actually helping the NE division in this type of analysis, though.
Puck Worlds: Chasing Pucks from here to Turku.
For Twitter Updates on Puck Worlds, follow @puckworlds. For updates plus additional witty banter from yours truly, follow @saskhab.
Yet Montreal finished (in EC): 7th, 10th, 1st, 8th, 8th.
Toronto: 9th, 9th, 12th, 12th, 15th
Buffalo: 3rd, 1st, 10th, 10th, 4th
Ottawa: 1st, 3rd, 7th, 11th, 5th
Boston: 13th, 13th, 8th, 1st, 6th
One could argue Montreal as being lucky over that time frame, but such ‘luck’ has probably hindered the team long term.
Puck Worlds: Chasing Pucks from here to Turku.
For Twitter Updates on Puck Worlds, follow @puckworlds. For updates plus additional witty banter from yours truly, follow @saskhab.
Of course we need to remember that Corsi numbers factor out goaltending ability and generally speaking the SE division has pretty bad goaltending (save for Florida) where as the NW has pretty good goaltending. That’s the problem with Corsi when attempting to evaluate teams. The fact is, only twice since the lockout has a NW team had a losing record against a SE team.
2009-10 Minnesota Wild (1-2-3)
2008-09 Calgary Flames (1-3-1)
Since the lockout the NW is 64-31-12 against the SE and the SE is 43-49-15 against the NW. Still think the difference between the NW and SE is negligible?
----------------------------------------------------------------------------------
HockeyAnalysis.com - Taking a Deeper Look at the World of Hockey
Every time I see someone write “that’s the problem with Corsi in [blah blah blah]” it usually reflects a misunderstanding of what makes Corsi so useful to begin with, i.e. a pretty good indicator of the more repeatable and sustainable and less luck-driven parts of hockey, that is gaining territorial advantage and outchancing.
By the way I have read your “analysis” of Corsi on your website and frankly it’s a load of shit.
by R O on Jun 9, 2010 2:37 PM EDT up reply actions
But it isn't...
This Corsi analysis says that the NW and SE divisions are essentially equal. Are you trying to tell me that the NW’s dominating record against the SE can be attributed to luck?
----------------------------------------------------------------------------------
HockeyAnalysis.com - Taking a Deeper Look at the World of Hockey
by HockeyAnalysis on Jun 9, 2010 3:12 PM EDT up reply actions
The Corsi analysis suggests the NW/SE divisions are about equal at controlling the play at even strength. Whatever other factors come into play (including goaltending performances) they are not nearly as dependable or controllable or attributable to skill as controlling even strength play.
by R O on Jun 9, 2010 3:29 PM EDT up reply actions
My comment below should be here.
----------------------------------------------------------------------------------
HockeyAnalysis.com - Taking a Deeper Look at the World of Hockey
by HockeyAnalysis on Jun 9, 2010 3:43 PM EDT up reply actions
So you agree that...
…this analysis isn’t actually telling us how good or bad the NW division is in relation to the SE division as far as actually winning hockey games is concerned, only that at even strength when the game is tied they seem control the play equally well (or equally poorly as the case may be).
----------------------------------------------------------------------------------
HockeyAnalysis.com - Taking a Deeper Look at the World of Hockey
I don’t agree with what you say. Even strength play is a huge component of winning games. Even more to the point, even strength play (or the range of reasonable expectations thereof) is an even bigger part of the aspect of winning that teams actually control over time-spans of 82 games.
Whereas the other major factors of goaltending or finishing ability, well the players certainly control the midpoint of the range of reasonable expecations but the range is so wide and the outcome within the range so luck-affected that predictive power goes right out the window.
Shit man it seems you want a magic bullet, and there is just not one.
by R O on Jun 9, 2010 4:01 PM EDT up reply actions
But...
This Corsi analysis does a horrific job at predicting what the NW divisions record would be against the SE division which to me makes this analysis garbage, for whatever reason you want to put on it.
----------------------------------------------------------------------------------
HockeyAnalysis.com - Taking a Deeper Look at the World of Hockey
by HockeyAnalysis on Jun 9, 2010 4:11 PM EDT up reply actions
Why did the Minnesota Twins let David Ortiz go for nothing? Because they focused on what he couldn’t do, not what he could do.
Corsi at even-strength, score tied predicts about 40% of winning on its own. Don’t Ortiz it because it doesn’t predict 100%.
Also, quoting records that includes OT and SO can be very misleading. And don’t forget luck: even if the NW and SE were equally matched, the odds of one of them having a 64-43 record over 107 games is >5%.
So the problem is you are using a 40% tool to draw (incorrect) conclusions about which divisions are better instead of using the 100% tool, actual won-loss records.
----------------------------------------------------------------------------------
HockeyAnalysis.com - Taking a Deeper Look at the World of Hockey
by HockeyAnalysis on Jun 9, 2010 4:37 PM EDT up reply actions
You keep using your W/L records and goal differentials to try and predict future winning. See how that turns out (hint: you’ll lose money).
by R O on Jun 9, 2010 4:56 PM EDT up reply actions
Two equally-matched teams have a 5% chance of posting a .600 winning percentage in 107 games. So the notion that won-loss records are a 100% indicator of a team’s ability is incorrect.
So there is a 95% chance that a team with a .600 winning percentage in 107 games is better than their opponent. Most people consider that statistically significant.
----------------------------------------------------------------------------------
HockeyAnalysis.com - Taking a Deeper Look at the World of Hockey
by HockeyAnalysis on Jun 9, 2010 5:08 PM EDT up reply actions
Your OT/SO numbers are skewing the results. The SE’s true winning percentage is .430 in 107 games. 18% of the time we can get an outcome that extreme by chance.
Goaltending, shooting, PP/PK. It all plays a role. But controlling the puck when the game is close is the single-most important factor in determining a team’s true talent.
If we found a team with a .430 winning percentage that had 50% of the shots, you’re free to bet that it’s going to continue being a .430 team. I’ll bet on .500.
So to be clear, in your own words, a won-loss record analysis would indicate that the NW is better than the SE at an 82% confidence level. At the same time and yet you feel fine concluding that the NW is only marginally better than the SE based on even strength when tied Corsi analysis. Just curious, what is the confidence level in that conclusion?
----------------------------------------------------------------------------------
HockeyAnalysis.com - Taking a Deeper Look at the World of Hockey
by HockeyAnalysis on Jun 9, 2010 5:53 PM EDT up reply actions
Won-loss record is such a poor indicator of team ability that nobody uses it as a predictor of future performance anymore. Goals for and against aren’t much better. Shot differential is the best of the bunch.
I welcome your efforts to find other methods for determining a team’s skill level. I’ll happily link to your blog posts and we can discuss the merits of your approach.
I am not trying to say won-loss records are perfect, but you shoot me down over and 18% chance I am wrong in saying the NW division is better than the SE but won’t or can’t put your own analysis to the same significance testing. Is it 82% or better?
----------------------------------------------------------------------------------
HockeyAnalysis.com - Taking a Deeper Look at the World of Hockey
by HockeyAnalysis on Jun 9, 2010 6:16 PM EDT up reply actions
Really? Wouldn’t that imply that your Corsi model is no better or worse than the won-loss model, just different?
Your Corsi analysis: 82% liklihood NW division is only marginally better than SE, if at all.
Win-loss analysis: 82% liklihood NW division is significantly better than SE division.
Not sure I can accept both those statements.
----------------------------------------------------------------------------------
HockeyAnalysis.com - Taking a Deeper Look at the World of Hockey
by HockeyAnalysis on Jun 9, 2010 9:47 PM EDT up reply actions
Gabe, this doesn’t make a lot of sense to me. Why would the two models have the same degree of confidence by definition?
by Scott Reynolds on Jun 9, 2010 10:44 PM EDT up reply actions
Maybe I misunderstood what was being asked – if you get a .570 winning percentage in 107 games, then there’s an 82% chance that your true Corsi% is >50%, regardless of your observed Corsi.
I think what he was saying was that a winning percentage of 57% implies an 82% chance that the winning team is actually superior. Your comment was that, based on their Corsi the two division are about even. The question, then, is given the observed Corsi between the two teams, how confident should we be that the two teams are, in fact, of an approximately equal skill level. I don’t know the answer but, “the same 82%” seems like an odd answer.
by Scott Reynolds on Jun 10, 2010 11:49 AM EDT up reply actions
Really?
Why is there an 82% chance that real Corsi% is >50%?
----------------------------------------------------------------------------------
HockeyAnalysis.com - Taking a Deeper Look at the World of Hockey
by HockeyAnalysis on Jun 10, 2010 11:53 AM EDT up reply actions
I think you’re conflating the EV and special teams stuff a bit, Gabe. Goaltending too, to a lesser extent.
Corsi% with the score close is a terrific predictor of future EV GF/GA (sans empty netters). When you filter out the binomial chance variation.
Abilities of EV save% and shooting% simply can’t exist at the levels that so many frequentist hockey analysts claim. The universe won’t allow it.
That’s why I think PDO numbers work so well. For e.g. if the SE has generous shot counters, the EVsave% may get bumped back to league average, but it comes at the expense of EVshooting% . It’s spooky in any case.
The other reason is that there seems to be a relationship between scoring bias and goaltender ability, coincidence or otherwise.
That’s why DocMyBrainsHurt’s thinks that scoring bias doesn’t exist (his ANOVA math will actually imply a scoring bias made of antimatter) … so pray I’m right, Gabe. Otherwise there is a tear in the fabric of the universe, and it won’t end well for any of us. :D
Thing is, I’ve never told you any of this before, yet you naturally gravitate to tied-score-corsi% and PDO numbers (or a variant). You have a marvelous knack for this stuff, Gabe.
I count nine inter-conference divisional matchups. An 18% event happening in one of nine matchups sounds perfectly reasonable, even likely in fact.
by R O on Jun 9, 2010 7:06 PM EDT up reply actions
I haven’t looked at the other inter-conference matchups, have you?
----------------------------------------------------------------------------------
HockeyAnalysis.com - Taking a Deeper Look at the World of Hockey
by HockeyAnalysis on Jun 9, 2010 9:48 PM EDT up reply actions
In addition to NW vs SE I also just looked at NW vs NE and NW vs ATL
NW vs NE 65-40-14
NW vs ATL 56-37-11
NW vs SE 64-31-12
But of course all three of those eastern conference divisions have > 50% even strength game tied Corsi%. Also, the greatest div vs div Corsi% is central vs NW at 53.4%. In the past 5 years though the central has a 251-184-65 record against the NW and the NW has a 249-213-38 record against the central division. That isn’t near the difference one might suspect from the difference in Corsi%.
All this indicates that there is a systematic error occuring with this method for evaluation teams/divisions such that it consistently underestimates the winning percentage of teams in the northwest division. I haven’t looked but I suspect there are similar systematic errors for other divisions too, probably at least with the SE division getting systematically higher Corsi% ratings than their win% would indicate.
----------------------------------------------------------------------------------
HockeyAnalysis.com - Taking a Deeper Look at the World of Hockey
by HockeyAnalysis on Jun 9, 2010 11:00 PM EDT up reply actions
Why did the Minnesota Twins let David Ortiz go for nothing?
Cuz he used up all their steroids? :)
by Passive Voice on Jun 9, 2010 10:09 PM EDT up reply actions
I would focus on your line of thinking before qualifying other analyses as “garbage”. For instance this beauty:
Defensive minded players are not necessarily on the ice to control the play, they are on the ice to not allow goals against most typically by the oppositions top offensive forwards. As mentioned above, one way to accomplish this is to go into a defensive shell and just not give up any quality scoring chances against. A player can have a sub-par Corsi number, but be doing his job perfectly well.
betrays your complete misunderstanding of how the game works.
by R O on Jun 9, 2010 4:59 PM EDT up reply actions
Right. Though it sounds plausible, once teams go into a defensive shell (and everybody except Toronto does it), they don’t show much ability to impact the quality of scoring chances against.
Yeah. And the volume doesn’t decrease as much as some head-in-sky thinkers would hope, either. The neutral zone is not an easy place to keep the puck.
The shell appears to only produce winning results if the team already has a lead. Nobody goes into a game deploying the shell and actually expects the midpoint of reasonable expectations to be a win – only the range is widened by playing like that.
by R O on Jun 9, 2010 5:13 PM EDT up reply actions
You clearly didn’t watch the Avs in the playoffs this season
The New Improved Avalanche. Now with Real Coaches!
Jibblescribbits: C'mon over and waste some time
by Jibblescribbits on Jun 9, 2010 5:14 PM EDT up reply actions
Or the Habs? :-D
Yeah teams will do the shell, but to widen the range of reasonable expecations (i.e. give luck a bigger role in the result). But in their heart of hearts the coaches know that in a million parallel universes, their strategy won’t produce W’s in more than half of them.
A winning strategy, it ain’t.
by R O on Jun 9, 2010 5:17 PM EDT up reply actions
I think teams who are inferior will go into a shell a bit and then hope for an odd-man rush/ some lucky break for a win, especially in the playoffs. It obviously won’t work over an 82 game season, but in a 7 game series it probably maximizes the odds an inferior team will win.
The New Improved Avalanche. Now with Real Coaches!
Jibblescribbits: C'mon over and waste some time
by Jibblescribbits on Jun 9, 2010 5:26 PM EDT up reply actions
At least I am not trying to say: Lets ignore goaltending ability, lets ignore power play ability, lets ignore penalty kill ability, lets ignore shooting ability, lets ignore a teams play whenever the game is not tied, etc. when drawing conclusions about which teams or divisions are better.
----------------------------------------------------------------------------------
HockeyAnalysis.com - Taking a Deeper Look at the World of Hockey
by HockeyAnalysis on Jun 9, 2010 5:18 PM EDT up reply actions
You say this like it’s a good thing. Do you really take all of these “abilities” (or rather, measures of them over short time spans) to be as actually ability-driven as outchancing and territorial advantage at even strength with the score tied?
by R O on Jun 9, 2010 5:20 PM EDT up reply actions
Short spans?
We are talking 5 season of data. Is that really a short span? With respect to NW vs SE there are 107 games, which isn’t a ton, but when you combine that with all the other evidence (i.e. the overall western conference record vs eastern conference or that SE division has only produced 8 playoff teams in 5 years and the NW division has produced 13) it just seems silly to claim that the NW and SE divisions are almost equal. To me, something just isn’t adding up right with your analysis.
----------------------------------------------------------------------------------
HockeyAnalysis.com - Taking a Deeper Look at the World of Hockey
by HockeyAnalysis on Jun 9, 2010 5:33 PM EDT up reply actions
I don’t really understand what the debate here is. Nobody is claiming that Corsi represents 100% of the game, it just happens to be a skill that can be quantified with relative certainty over short time spans. Ironically, given that we’re discussing a span of 5 years here, we actually have enough data that we could figure out if a division had an advantage in other skills. Goal differential over 5 years is not that noisy; however, it’s only useful in retrospect, since the teams aren’t static over that time and are certainly no longer the same now.
by Tom Awad on Jun 9, 2010 8:11 PM EDT reply actions 1 recs
I showed above that NW vs NE and NW vs ATL had similar issues to what has been pointed out about NW vs SE and even NW did better vs Central than their Corsi% would indicate they should. Do they have similar shot% differences vs NW compared to overall? I suspect so.
----------------------------------------------------------------------------------
HockeyAnalysis.com - Taking a Deeper Look at the World of Hockey
by HockeyAnalysis on Jun 9, 2010 11:09 PM EDT up reply actions
That is the point I am trying to make...
Corsi% applied in this manner is pointless because there are far better methods for evaluating which teams/divisions have been the best over the past 5 seasons, starting with won-loss records but certainly goal differential and probably other methods can be considered as well. I think it can be shown (if I haven’t already) that Corsi% fails fairly significantly in this retroactive 5-year analysis.
----------------------------------------------------------------------------------
HockeyAnalysis.com - Taking a Deeper Look at the World of Hockey
by HockeyAnalysis on Jun 9, 2010 11:05 PM EDT up reply actions
Comments For This Post Are Closed
Secondary Sidebar
User Tools
FanPosts
Community blog posts and discussion.
Certain photos copyright © 2012 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
