• CorsiRel And Famous Defencemen I

    by  • May 30, 2014 • Hockey • 21 Comments

    I wrote about evaluating defencemen using CorsiRel the other day, making an argument that, generally speaking, a defenceman’s 5v5 contribution is equal to his impact on Corsi%. Now, that comes with all sorts of caveats. Not all minutes are created equal – it’s tougher to play against Sidney Crosby than it is against Luke Gazdic. Some guys get harder zone starts than others. Some guys are on better teams than others. There are a lot of knots to unravel in order to get to the point of precisely quantifying this. Theoretically, I’m pretty sure that I’m right in the vast majority of cases though.

    Unfortunately, nobody’s really presenting CorsiRel for more than a few years – Extra Skater has it but stats.hockeyanalysis.com doesn’t. In light of that, I put together a series of graphs of famous defencemen and how they’ve done in terms of CorsiRel since 2007-08. I wanted to mitigate the impact of who guys play against a little bit, so I only included seasons in which a guy played at least 41 games and at least 19 minutes a night. Ideally, I’m capturing top four defencemen. I then went through and graphed a team’s Corsi% when the defenceman in question wasn’t on the ice and when he was. I think it’s kind of illuminating to see this laid out this way.

    For technical reasons (ie. to simplify the work), I’ve also ignored seasons in which a guy played for multiple teams. Pay attention to the years at the bottom of each graph. The red line in each graph represents the player’s Corsi; the blue line is the team without the player on the ice.

    Screen Shot 2014-05-30 at 2.18.12 AM

    One of the things that struck me as I went through this data is that Aucoin put up decent numbers pretty much every season for which we have lots of data. Above 50% every year and generally better than his teams.

    Screen Shot 2014-05-30 at 2.18.24 AM

    That’s something that more famous defencemen can’t say. For a guy who’s made a lot of money in the NHL and garnered a lot of fame, it puzzles me that Bouwmeester’s never been able to hit 50% while outperforming his team until he was paired with Alex Pietrangelo. This will come up again but if a guy’s an elite defender, shouldn’t he be able to out-Corsi his team without a Norris-calibre partner?

    Screen Shot 2014-05-30 at 2.18.32 AM

    For a guy who’s a soft puck mover, Campbell seems to do nothing but post Corsis a few points higher than his team, even if his team is really good, as the Blackhawks were when he played there. Amazing that he kind of slid through the league and ended up in Florida when Chicago needed to trade him.

    Screen Shot 2014-05-30 at 2.18.39 AM

    Does Chara have any competition for best defenceman of his generation? Those are killer numbers, as he comfortably outperforms a pretty good team. What’s more, we know that he wasn’t being sheltered or anything. If you think of a generation as being five years before and after a player’s draft, the only players I can see in the discussion are Scott Niedermayer (1991) and Chris Pronger (1993). After that, well…Drew Doughty and Alex Pietrangelo in 2008?

    The amazing thing, as I’ve mentioned before, is that Chara was probably having a huge possession impact on the Sens in 2006, when they chose Wade Redden over him and saw their possession numbers go off a cliff. An incredible player who probably doesn’t get the credit for being an incredible player that he deserves because people are so focused on the size thing.

    Screen Shot 2014-05-30 at 2.18.53 AM

    Speaking of Drew Doughty…these are pretty wild numbers too. Better than his team every year since his rookie year and breaking 58%. Those are incredible numbers for a guy matched up against the opposition’s best.

    Screen Shot 2014-05-30 at 2.18.59 AM

    I’ve included Edler just because he took a lot of flak this year. A guy who can tread water on a 52% Corsi% team is a guy who can, at the very least, be a second pairing guy on a Stanley Cup contender. Vancouver’s got problems but I’m not sure their defence is really on that list.

    Screen Shot 2014-05-30 at 2.19.05 AM

    Note how Ehrhoff has pretty much always outperformed his teams. He did suffer a sharp decline this year relative to his team, which is interesting to me because I vaguely recall that his numbers were pretty solid early – my sense is that he must have fallen off as the year went along. Whether ennui or decline will be seen, I guess.

    Screen Shot 2014-05-30 at 2.19.13 AM

    I know it’s cool to rave about Oliver Ekman-Larsson but the Yotes weren’t a great team this year and he still couldn’t beat their Corsi%. What’s more, this is the second year it’s declined relative to his team. Worth keeping an eye on.

    Screen Shot 2014-05-30 at 2.19.22 AM

    Giordano really did have a bonkers season. I’d kill to have SportVu data for a couple years for Calgary so I could try and figure out what was different. He seems awfully old to have made such a leap. Note that he consistently beat or matched the team in previous years though, something that Bouwmeester didn’t do.

    Screen Shot 2014-05-30 at 2.19.31 AM

    Victor Hedman is starting to look an awful lot like a superstar. With a healthy Steven Stamkos next year, Martin St. Louis’ salary room and a couple of high draft picks, Tampa looks to me like they might be a team the Oilers face in a Stanley Cup final at some point.


    Screen Shot 2014-05-30 at 2.19.40 AM

    Analytics punchline.

    Screen Shot 2014-05-30 at 2.19.47 AM

    Makes his team much better when he’s on the ice. Ken Hitchcock has sniffed that he doesn’t play like a traditional defenceman but does that matter? The name of the game is doing things to make your team better.

    Screen Shot 2014-05-30 at 2.19.54 AM

    Keith’s name came up when I inquired on Twitter as to a defenceman drafted since Chara who might plausibly have a better career. I don’t really see it – he’s done well on a great team and been better than his team most years but he doesn’t seem to me to quite reach Chara’s heights, particularly when you consider that Chara’s doing what he’s doing in what should be the twilight of his career.

    Screen Shot 2014-05-30 at 2.20.02 AM

    The King. I don’t think history will look too kindly on that 2010-11 Norris Trophy – he seems to have done a lot worse than his team that year – but his numbers before and after were excellent. The voting in 2010-11 was extremely close, with Weber and Chara both finishing within 50 points of Lidstrom, despite Lidstrom outscoring them by 14 and 18 points respectively. Given what we’ve learned about PHWA practices, it would be interesting to see the votes from that season and get a sense of how things broke down between voters you’ve heard of and voters you haven’t.

    I’ll do the other half of the league early next week.

    Email Tyler Dellow at tyler@mc79hockey.com


    21 Responses to CorsiRel And Famous Defencemen I

    1. D. Green
      May 30, 2014 at

      Interesting – just wondering – OEL – increased responsibility ? And Jack Johnson – is there an improvement going on (relatively speaking that is) ?

      • Tyler Dellow
        May 30, 2014 at

        Yeah, both of those things are probably true. At the same team, OEL getting beaten up by better competition doesn’t necessarily help Phoenix. And Johnson is 27. Nice that he’s stemmed the bleeding sort of but he’s kind of out of time.

        • TruthObserver
          May 30, 2014 at

          He’s also only 22 years old. Most defensemen are just starting to enter the league at OEL’s age. Still lots of time for improvement.

    2. Pingback: Spectors Hockey | NHL Blog Beat – May 30, 2014

    3. May 30, 2014 at

      Fun to see this stuff in such an understandable visual form. Thanks for that.

      With regard to not having the data for more than a couple of years, I think Behind the Net has RelCorsi data from 2007-08 through 2013-14.

      With regard to specific players, you’ve mentioned it already, but it seems like ZoneStarts are going to have a large impact. There have been some different attempts to quantify that impact, including your own use of “Open Play”, which doesn’t so much correct data as exclude some of it, but that has the advantage of sidestepping some of the debate about how much a DZ start is worth in terms of Corsi and that’s quite nice. I know that there are other factors to consider like QT and QC and % of the time played with the lead v. tied v. trailing, but it seems like we’re reasonably well-equipped to account for the difference in zone starts (especially if you’ve got Open Play data going back several years), so I was wondering why you decided against taking it into account here.

      A couple of examples of the possible impact would be looking at a guy like Doughty (who has consistently been over 50% in the OZ and middle-of-the-pack on his team) who would appear worse if we corrected in this way, and a guy like Chara (with a lot of DZ starts relative to his teammates and less than 50% each year) who would appear even better than he already does in your analysis above.

      • Tyler Dellow
        May 30, 2014 at

        Scott, that’s a good idea. Honestly, just did it this way because it was easier to assemble the data. I’ll try the open play way sometime this summer.

    4. Robert
      May 30, 2014 at

      I agree with Scott. So easy to understanding the visuals.
      A 57% Corsi at 42 years old? Wow… Makes you wish you could see Lidstrom’s numbers from earlier in his career.

    5. Stretch
      May 30, 2014 at

      Thanks for including Aucoin. I liked him a lot when he was with the Islanders and was disappointed he wasn’t kept. He is one of the most uderrated defensemen of the last 15 years.

    6. josh
      May 30, 2014 at

      51% as his worst year. Fuckin Lidstrom

    7. StatsGoon
      May 30, 2014 at

      Awesome stuff.

      How much would you love to have the data on Orr?

      With Ehrhoff, it looks more like some regression from the insane 8%+ CorsiRel he had from 2013 moreso than anything. I’m sure ennui definitely applies at a certain point as well.

    8. Aaron
      May 30, 2014 at

      I would love to see Lidstrom’s earlier numbers. Jesus, 60% in 2008? Thats ridiculous. And yeah, 2011 doesnt loom great until you realize its still 51% and likely the worst season of his career.

    9. Geoff
      May 30, 2014 at

      Any thought to doing Letang? Was he left off because of his injuries? He strikes me as similar to Karlsson in that he has a bad reputation for Defense but generally drives play in a positive manner. On the other side of that coin, would love to see Mike Green, a D-man who probably deservedly gets lumped in as a “4th forward” and hurts his team on D.

      • Leigh
        May 30, 2014 at

        I have the opposite reaction to you.

        I’d love to see how Green fares. Because he was the only defensemen on the Capitals this past season with decent numbers for possession–on a team that was one of the worst possession-wise under Oates.

    10. Bank Shot
      May 30, 2014 at

      How much more predictive is relCOrsi than just going off the boxcars?

      If I just take a sample of the top 20 defencemen in scoring I would say the vast majority are bound to have high relCORSI’s.

      There are going to be exceptions like Jack Johnson, but there will be RELcorsi exceptions as well in the famous D man list. Pronger’s graph won’t look as good as Erhoff’s for example. Ryan Suter will not look great since he joined the Wild.

      Guys like Hjalmarsson and Mitchell will never look good by RelCorsi but they are still good players.

      Brian Campbell has been given the best available zone starts on his team every single year with the exception of 2013-2014.

      If you believe zone starts matter than you have to find a way to adjust RELcorsi to level the playing field or you don’t have a stat that is the most meaningful metric IMO.

      Top ten in 2013/14 in boxcars:


      Top ten in RELCorsi(playing more then 15 ES min per night):

      Boxcars is more predictive of being a good defenseman if you just take it at face value in this instance as the first group is clearly stronger.

      If you start taking zone starts, quality of comp/teammates into account, then you start getting more and more subjective as there is no single standard to weigh these factors.

    11. Sapp
      May 30, 2014 at

      This is a friendly reminder Lidstrom was 41 years old for the 2011-2012 season. That’s incredible.

    12. Pingback: The Kesler Effect | The 6th Sens

    13. Pingback: The Fancy Stats Thread - Page 22 - SportsHoopla.com Sports Forums

    14. Sven22
      June 2, 2014 at

      For some context on Lidstrom’s one down year relative to the rest of the Red Wings:

      1) He was anchored to Brad Stuart basically all year (shared 60% of his even-strength ice time) and spent about half the remainder with Kronwall, who has himself (despite his other strengths) never been much of a needle mover at 5-on-5.

      2) He actually had tough zone starts (less than 49%) that year, quite possibly to spare his offensively gifted but defensively challenged ex-partner Brian Rafalski (54%) from tough minutes.

      3) His quality of competition was nuts that year (even for him). It’s too bad Extra Skater doesn’t go back far enough to show a deployment chart, but Lidstrom’s Corsi Rel QoC was tops among NHL defenders by a light year in 2010-11.

      3) He was 40 years old — actually he turned 41 between the first and second round of the playoffs.

      Basically, Babcock decided to use Lidstrom as a defensive specialist at even strength and then let him loose on the powerplay. Which, incidentally, he dominated (39 powerplay points vs. 31 for next-best Visnovsky).

      And he still was at about 51% Corsi at even strength, against possibly the best competition in the entire NHL, playing with Brad Effing Stuart, starting less than 49% of his non-neutral draws in offensive zone, at age 40.

      He still probably didn’t deserve the Norris, granted, but given the circumstances I think his numbers are lot more impressive then they might initially seem.

      • John
        June 2, 2014 at

        This actually makes me cry that we don’t get to watch this guy play anymore. He posted a 60% CF at age 38? What in the actual ****.

    15. Pingback: Wie gut sind die Schweizer Verteidiger wirklich? | Emptynet Hockey Blog

    16. Pingback: Context Neutral Player Evaluation: Examining Defence and Calculating xGD20 | puck++

    Leave a Reply

    Your email address will not be published. Required fields are marked *