that was… well, really interesting to me actually. curious to hear more about it, mainly due to the strange flux in teams like Philly and the Rangers. it’s actually kind of weird.

Posted by perfection from LaLaLand on 07/17/09 at 03:00 AM ET

Seems to me at first glance that Corsi numbers as a team stat is something of a very weak predictor of success by themselves.  It would be easy to assume that a team with a Corsi number 200 points higher than the next highest team would be a playoff team.  However, with only 11 of the top 15 teams taking up the 16 playoff spots, it seems that the number should be higher.  If Corsi numbers were a strong predictor, I would expect no more than two top-15 teams to have missed the playoffs.  Meanwhile, team +/- meets this criteria with only 2 of the top 15 having missed the playoffs (strong, but not very strong)

What Corsi Numbers and +/- combined seems to indicate is how strongly a team’s defense/goaltending plays.  Detroit is #1 in Corsi and #5 in +/-.  The number 4 +/- team, Vancouver has a Corsi number over 1,000 points lower.  The number one +/- team has double the goal differential but falls more than 850 Corsi points lower. 

Of course, when you look at what these numbers mean, the findings are obvious.  If you have a high shot differential, but let in more shots than the opposition (through weaker goaltending/defense), then your goal differential is going to suffer. 

What I would like to see is the Corsi number for Pittsburgh since Bylsma took over as head coach run out over the course of an 82-game schedule and see what that would do to their ranking.  Hindsight being 20/20 it’s perfectly clear to see that Therrien was not running that club the way it should have been run and I think this stat suffers from it.

Posted by J.J. from Kansas on 07/17/09 at 08:28 AM ET

Sorry, meant to say “weak” instead of “very weak” predictor in the first sentence.

Posted by J.J. from Kansas on 07/17/09 at 08:29 AM ET

JJ—don’t forget that Corsi is purely an even-strength measure. It has no bearing on success on special teams and that may well be an important consideration on

Which brings me to a significant oversight in the chart. The +/- and Corsi being compared here are really apples and oranges: Corsi is measured purely at even-strength. But plus-minus isn’t, as while it discounts power play goals, it doesn’t discount shorthanded goals.

It’s a significant difference in some cases. Why does Philly have a poor Corsi and a good +/-? A lot if it is because they have 15 shorthanded goals for and 1 shorthanded goal against. That corresponds to a +10 even-strength plus/minus—which ought to push them down a fair bit down the chart. Conversely the Rangers allowed 13 SHG and scored only 7; -14 plus-minus would bring them up the rankings a bit.

Ideally a chart like this would also discount empty-net goals, too, but that’s not nearly as easy to figure out just from reading the nhl.com stats because they don’t specify the game situation of EN goals.

Posted by MathMan on 07/17/09 at 10:25 AM ET

“JJ—don’t forget that Corsi is purely an even-strength measure. It has no bearing on success on special teams and that may well be an important consideration on “... team success as measured in the standings.

Sorry about that.

Posted by MathMan on 07/17/09 at 10:26 AM ET

“those who put puck in net the most in the correct circumstances win the cup”

thats the only stat you need. this stuff has absolutely no context to equate them to other than to create abstraction. seeing the Penguins at 21 for the season obviously has NO contextual significance.

this stuff might be useful to generate “trends” over periods but it tells you nothing about the actual context - therefore its practically useless. theyre stating the obvious, “the teams with better CORSI’s are on average good teams.” no crap? eh? we needed all those stats to tell us that instead of just looking at the standings?

trying to apply this baseball kind of junk to a game so chaotic as hockey is a real disappointment. hockey should be cherished for its constant chaos, unpredictability and immeasurable events.

Posted by Death Metal Nightmare from MKE on 07/17/09 at 12:48 PM ET

MathMan, I was looking at the stats for goal differentials and how they differ from team +/- and it looks like those do create a bit of difference.  However, in Philadelphia’s case (+24 team plus/minus with a plus-28 goal differential), it looks like their defense generally found a way to give back any shorthanded goal bonus they got.  The Rangers goal differential was -12, an 8 point improvement over their team plus/minus.

The Sharks topped the league with a 36 point difference between their goal differential and their team +/-  St. Louis was the exact same number in the negative to bottom out the league. 

Overall, it does look like goal differential itself is a more important stat than team +/-.  While only 12 of the top 15 teams in that category made the playoffs, no playoff teams were ranked below #19 (NYR).  The three non-playoff teams that snuck into the top 15 were MIN at #11 with a plus-17 differential, BUF at 13 and +13, and FLA. at 14 and +8.  Those three teams’ Corsi numbers are all negative.  If I had to make sense out of this jumble, I would say that these stats indicate that there’s a bigger separation between the best teams and the middle-ground teams than the league wants with their “forced parity” agenda.  Boston, San Jose, Chicago, Detroit, and New Jersey are the top five goal differential teams.  Four of them are in the top five in team +/- (SJ falls to #10).  New Jersey and Boston fall out of the top five in Corsi, but are still top-half. 

Vancouver is close in team plus/minus and goal differential, but their negative Corsi shows what this playoff year seems to have indicated as well, they’re a decent team that is too reliant on their goaltending.

As an addendum and an olive branch to the Pittsburgh fans out there.  Let me reiterate that I am very interested to see what the numbers would be if Bylsma had started the year as the coach.  I truly believe Therrien damaged their stats to the point where he took them out of the running for those top five teams stats-wise over the course of the regular season.  The Penguins proved that they belong among the elite teams in the NHL.

Posted by J.J. from Kansas on 07/17/09 at 01:06 PM ET

Paralysis by analysis. If you just look at the overall +/- (es,pp & sh situations) you get a very good idea of who’s playing good. I beleive the top 7 of 8 in each conference made the playoffs. Maybe 1 or 2 get in the playoffs with a bad +/- as compared to teams around them in the overall standings. Overall +/- ,simple but accurate and very telling.

Posted by Lindas1st on 07/17/09 at 01:24 PM ET

Simply put Total Goal Differential is a key indicator.

Posted by Lindas1st on 07/17/09 at 01:34 PM ET

Goal differential is about as good a yardstick of team strength as there is; it will be a rare case where goal differential doesn’t predict standings points, and those teams are typically called “lucky” or “unlucky”.

That said, in order to evaluate team strengths and weaknesses it’s useful to point out that there are three kinds of “goal differential” here and that we should be clear which one is being used. Pure goal differential includes all goals (including all special teams goals); plus/minus discounts power play goals for and against; and even-strength goal-differential only considers even-strength goals and not special teams goals.

For Corsi, ES goal-differential is what matters as it is a measure of shots attempted at even-strength. A useful statistic to look over on the NHL website is the 5 on 5 goal ration (abbreviated “5/5 F/A”)—it’s 5-on-5 goals for divided by goals against and indicates how successful a team has been at even strength.

Corsi is generally thought to be a good measure of puck possession. It’s helps measure how good teams are at controlling the play. It directly correlates with ES scoring for and against, but while 5-on-5 play is vitally important it’s not the only factor in a game. A team could make up for mediocre even-strength play with a dominant power play, for example (Philadelphia, for example, has been average to poor at even-strength for the last couple of years, but has had a strong power play; Montreal is very similar). That’s siginficant information at telling us WHY a team is good, or not. And how reproducible those results might be, year over year.

For example, this should give some pause to people who think that Boston is “for real”. They’re really not all that good at controlling the flow of play 5-on-5, and their penalty kill is middling. In other words, they’re not nearly as dominant as their goal differential and standings would indicate. Without utterly mind-blowing goaltending from Tim Thomas, they wouldn’t look nearly this good, though a very strong power play would keep them in an evniable spot. Unless you think that Tim Thomas can consistently put out .933 save percentage seasons (in which case he belongs in the HoF and in best-goalie-ever discussions), it’s likely Boston will take a step back—even if Thomas reverts to merely excellent it will be a significant drop-off.

Posted by MathMan on 07/17/09 at 02:51 PM ET

MathMan, great point.  Also, I agree 100% with your analysis of the Bruins and Thomas.

Posted by Lindas1st on 07/17/09 at 02:59 PM ET

Giving this a cursory glance, and admitting I haven’t read up enough on this, this almost looks like a rating of team minus goaltending.  In general, it looks like teams that rated a higher Corsi number than finish had poor goaltending, and a higher finish than Corsi better quality goaltending.

Posted by Wags` on 07/24/09 at 01:58 AM ET