• The Bigger Picture: Oilers, Possession and Regression

    by  • March 27, 2013 • Hockey • 13 Comments

    I have a bit of a personal policy that I don’t criticize hockey teams for making decisions that I agreed with at the time if things subsequently go bad. Any moron can point out when something doesn’t work and if I sat down at the time and figured it was a good move, it’s not really fair for me to dump on the team when it doesn’t work out. Hindsight’s 20/20 and all that.

    When you talk about what’s gone wrong this year for the Oilers, the biggest thing has been the inability to score goals at ES. I made this graph a few days ago, so events will have surpassed it a little bit with Eberle/Hall/RNH having a big game at St. Louis last night but it illustrates the problem pretty clearly. It’s a comparison of 5v5 shooting percentages last year versus this year for returning Oiler forwards:

    So, if you’re looking for positives, Belanger and Petrell, two of the most irrelevant pieces of the Oilers’ offence are enjoying bounce back seasons in on-ice shooting percentage. Unfortunately, they take a tiny share of the Oilers’ 5v5 shots. Meanwhile, the guys who’re supposed to score the goals have been suffering through miserable years. It’s unfortunate but it’s a risk of a Small Sample Season.

    Just to expand on that a little bit, assume that Jordan Eberle is a guy who has a true talent 5v5 on-ice shooting percentage of 9.7%, which is fantastic. He’s currently got an on-ice shooting percentage at 5v5 of 6.55%. Assume further that he’ll be on the ice for 360 shots in a 48 game season (15 minutes a night multiplied by 48 games multiplied by 30 S/60). The probability of him having an on-ice S% of 6.55% or worse over that time probability 2.7% – it’s a small chance but not so small that we wouldn’t expect someone to get burned. If you expand that to 615 shots, or an 82 game season with the same assumptions, the chances becomes 0.3%. So far this year, Eberle’s been on the ice for 275 shots at 5v5 and 18 goals have been scored. The probability of that, or something worse, happening, if he’s a 9.7% true talent guy, is 4.2%. Call it a 24/1 longshot. It’s a longshot, but not so long that you don’t expect it to happen from time to time.

    In a longer season, a lot of that sort of stuff washes out from the data – not all of it, but a lot of it. This year, there’s just too little time for that to happen. So the Maple Leafs will probably go to the playoffs and the Oilers almost certainly won’t and that’s hockey. Of note: some other Oilers sites are discussing how the Oilers are going to need to trade away some of their skill for size today because they aren’t big enough and all that. I’ve heard such august panels as the HNIC panel allude to the fact that the Oilers don’t get to the net enough and take easy perimeter shots. I wonder where all these guys were last year, when everything that Eberle and, to a lesser extent, RNH touched ended up in the net. Did the Oilers get less big and tough over the long summer?

    Anyway, my point is that at least part of what’s gone against the Oilers this year has been the hockey gods not favouring them and if you aren’t willing to live with that possibility, you shouldn’t be a hockey fan – get into figure skating or something, where the gods are sitting beside the rink and open to deals. Part of it has been the predictable implosion of the defence corps and terribleness of the motley collection of fourth liners. This was easily foreseeable and ought to be chiselled on the grave of the career for whoever takes the fall if and when the Oilers miss the playoffs.

    The real mystery though, the thing that I don’t think anyone could have seen coming, is just how bad Sam Gagner and Ales Hemsky have done possession wise at ES. I won’t blame the Oilers for not foreseeing this and it’s not generally a risk that just comes with running a hockey team – guys who can put up good possession numbers don’t tend to just go completely in the toilet.


    I’ve pulled together some data from Hemsky and Gagner over the past six years. This isn’t with just each other, it’s overall.

    The final column on the right is the share of Oiler shots that these guys took while they were on the ice. As you can see, there’s something unusual going on. I tend to think that a players’ shot volume will be pretty tightly related to the number of shots that his team takes. Hockey’s a team flow game and it’s not like shots are generated by the player shooting the puck alone. Some players have a tendency to shoot more than others, but my mental framework for this is that as a team has more possession and generates more shots, a rising tide will lift all boats. If you ignore this year for Hemsky and Gagner, you can see that that’s just about entirely true: other than Hemsky having one year in which he took 6.8 S/60 while the Oilers generated 28.4 S/60 and a year in which he took 6.9 S/60 while the Oilers generated 28.1 S/60, their shooting rates rise with the team’s shooting rates.

    This year though, that’s just completely not the case. Pre shoulder surgery in 2009-10, Hemsky took about 8.0 S/60 at 5v5. He’s always had a bit of a bum rap as a guy who doesn’t shoot the puck – it was true when he was younger but he kept trending up and up until 09-10. He wrecks his shoulder in November, has surgery, comes back and suffers through another partial season, leading to a surgery on the other shoulder and then didn’t really look like right until about halfway through last season. There’s no easy breakdown available but his shot totals were higher post-ASB, which meshes with what I remember.

    Gagner’s path is a little different – he had a big Corsi/individual shots year in 2009-10, which is funny given how terrible that Oilers team was but he was kind of moved around the lineup a bit that year and (I’m guessing – I’ve blacked most of it out) maybe got a lot more time against bottom six players. He’s having a pretty good shot rate year this year by his own standards though. What’s kind of bizarre is that they both seem to be generating a lot of shots by their own standards but that this rising tide has lifted no other boats when they’re on the ice – Gagner’s taken 27.1% of the Oilers’ shots when he’s on the ice, besting all but his 2009-10 season. Hemsky’s taken 30.4%, which is crushing his previous high of 25.4%.

    If you go back to my theory about the volume of shots you take being a function of your own willingness to shoot when presented with the opportunity to do so and the number of opportunities with which you’re presented by virtue of your team being in the offensive zone, you wouldn’t expect things like this to happen. There’s something going on here that the data we have available to us doesn’t capture, I think.

    My second point is to do with the SF+SA/60 when Gagner or Hemsky are on the ice. Again, the point I’m going to make is more pronounced with Hemsky. The Oilers are only generating 0.5 S/60 fewer this year than last year when Gagner is on the ice but are allowing an extra 3.6 SA/60. In Hemsky’s case, they’re down 1.2 S/60 while allowing an extra 6.7 SA/60. Both guys have seen their SF+SA/60 explode when they’re on the ice; in both cases, it’s mostly been shots against.

    Again, I think that there’s something unusual going on here, although I’m not entirely sure what. It strikes me as odd that the SA/60 would increase by so much more than the SF/60 went down when neither guy is really moving up in the lineup – there tend to be more total shots per 60 minutes when guys in the top of the lineup are on the ice than when fourth liners are out there – but neither Hemsky nor Gagner made a big leap up the lineup this year.

    If you back out the goals involving an empty net at either end, Hemsky’s EV+13 and EV-15 while Gagner’s EV+15 and EV-19. They’ve been together for about half the season so the Oilers are probably something like -4 or -5 with one of them on the ice. With the position that the Oilers are in, a goal against here or there might have made a huge difference. It’s worth pointing out that the abysmal Corsi% of Gagner and Hemsky has hurt the Oilers this year but I’m not yet convinced that it’s happening because of something that they’re doing or not doing, given that they’re each individually generating a lot of shots and that there’s been a sort of ballooning of the SA/60.

    I know, from the comments here and that people make to me elsewhere, that I’m kind of perceived as a guy who thinks you throw scouting out the window and rely on data alone. This, I think is a perfect example of a case in which both scouting would help as well as better data. A team that tracked every touch of the puck over the years as well as as how much time tended to be spent in each zone would be extremely well positioned to figure out precisely what’s happening with Gagner and Hemsky and why their numbers have changed in such curious ways. Similarly, a scout with a day to spend in a video library might be able to come up with something.

    This question of why is extremely important for the Oilers because it’ll go a long way to determining the future shape of the team. If Gagner/Hemsky/? can be in the black in terms of Corsi% in a second line role, the Oilers have a really good second line, because those guys can finish. Hemsky’s got a long history of respectable Corsi results; Gagner less so, although he’s done well with Hemsky historically. It strongly seems to me like there’s something going on with Hemsky/Gagner’s Corsi% that I can’t figure out. You kind of worry though, that they’re running out of time to fix it with the Oilers and that they’ll be sacrificed to the demands for MOAR SIZE that you here emanating from certain corners of Oiler fandom.

    Email Tyler Dellow at tyler@mc79hockey.com


    13 Responses to The Bigger Picture: Oilers, Possession and Regression

    1. Ryan V
      March 27, 2013 at

      I like your theory of a player’s shot total is roughly willingness-to-shoot multiplied by time-with-possession. It gets tricky if you accept that willingness to shoot affects how much time your team has possession. Taken to the extreme, imagine a team that shoots the puck every time they cross the red line–by shooting so soon, they have very little chance of recovering the puck and are basically giving up possession.

      Now suppose that Hemsky and Gagner had a roughly optimal willingness-to-shoot over previous years, but that for whatever reason, they decided to shoot more often this year. If you buy the argument above, I think you’d see three things in the data:
      1. The total SF+SA while they’re on the ice would increase (check)
      2. Hemsky and Gagner would account for a larger share of their team’s shots when they’re on the ice (check)
      3. The Oilers’ share of shots with Hemsky and Gagner on the ice would go down, because they’re less likely to recover the puck after their shots (check)

      I don’t watch the Oilers nearly enough to say whether that’s what’s going on, but it would fit the data.

      • Tyler Dellow
        March 28, 2013 at

        Yeah, it’s not a bad theory. Anecdotally, I might guess that Gagner/Hemsky have just been shooting a lot because they have no other play a lot of the time. Neither guy is a selfish player but, for example, watching the Oilers play the Blues. I was struck by how little there was in terms of options for a pass.

        • Latif
          May 31, 2013 at

          caili0921They thought they were on to sontehimg afterplaying a brilliant game Friday night againstColorado, and had visions of making it two in a rowSaturday against the Flames.GD Star Ratingloading…

    2. chartleys
      March 27, 2013 at

      Wouldn’t some of this simply be due to usage? Running it through it by my eye, at least early in the year, it seemed like that line was getting a lot more D zone starts. This likely continued with Horc out as well. Pushing the puck out of his own end still isn’t a skill I’m overly impressed with Gags on. I still almost feel like his coverage seems like he would be more suitable on the wing than the center. Offensive zone he can play the center role pretty well, but defensively I still don’t like what I see. Step up in responsibility for gags, an off wing rookie (I know Yak’s numbers looked better than these two, by these metrics, but that line just didn’t look right together)

      Not convinced of these things and am willing to accept that it’s just my eye looking for errors when Gags is defending and not noting good plays…(call it a minor case of the “Tom Gilbert” effect)

    3. David Staples
      March 28, 2013 at

      Interesting post, Tyler.

      Hemsky and Gagner have played against weaker comp than the RNH line, so this is all the more troubling.

      The main issue is both players are inconsistent on defence and generally weak defensively, IMO. Hemsky relied on Horcoff (and Penner) to do the heavy lifting on defence for years. Perhaps his better shots plus-minus in other years reflects that.

      Gagner has never been strong defence, though showed a bit of improvement last season, before becoming a sieve again at the start of this season.

      In basketball terms, these are two all-offence, no-defence players. Put them together with Yakupov, another weak defensive player, and you’re going to have major and constant defensive issues.

      I’m not sure Hemsky, as thrilling as he is on his runs up ice, fits with this team any more. He is a solo-ist who needs to play with support players, such as Horcoff and Penner, guys will subvert their own offensive games to some extent to play off of him, as he dominates the puck (Tarasov wrote about this kind of player in detail in Russian Hockey Secrets in reference to Bobrov).

      The support players must also cover for the solo-ist defensively to some extent, though Hemsky does have real value in advancing the puck up ice and keeping it in the offensive end.

      As for Gagner, he is a puck-watcher on defence still — it’s hard to change a young lifetime of habit fixated on the puck — but he did improve last year, he has had a better run of games on defence as of late, and I’ve noticed an increasing willingness to make tough, physical defensive players in his own end. We saw the same thing with Cogliano in his last year or two in Edmonton — more willingness to play tough in his own end.

      Gagner might be able to turn the corner, but I’d hate to see the Oilers pay big bucks to a kid who you can’t use against tough comp.

      • Bebe
        May 31, 2013 at

        SLY er en av de sve6rt popule6re kvinners meekrvarer, designere Ueda Mizuki (Ueda Mi ず ki), den opprinnelige for Moussy designer, etter e5 ha lagt ved Moussy uavhengig av SLY merkevare sf8t og krydret vill stil,GD Star Ratingloading…

    4. David Staples
      March 28, 2013 at

      When I say “big bucks” in regards to Gagner, I mean anything more than $5 million per year. Best case is to get him for around $4.5 million, at least if he stays hot, and given how GMs pay for points.

    5. Woodguy
      March 30, 2013 at

      I have no proof, but by eye the Oilers up ice pressure system where the D steps up and the Fs cover has not worked well.

      F’s don’t cover when they should, D making bad decisions on when to step up.

      I think that system is sound, but both the players (moreso the D than F imo) need to implement it better.

      It would also help to have better players. To run that type of system requires everyone knowing exactly where everyone else is supposed to go and actually executing it (see Jones lollygagging everywhere on the ice creating holes where they are not supposed to be)

      So when I see the Oilers getting more lit up in terms of SA/60 and then by eye I see way more odd man rushes, which often turn into multiple shots against, I think I can see why the corsi is in the tank.

      4-93-14 are just so goddam good together it doesn’t show up in their numbers, but its show up everywhere else.

      Petry was a rock last year and this year he looks hesitant and unsure when to step up and when to skate backwards and mind his gap.

      I really do think a lot of the SA is due to the system with poor implementation and meh players.

      An upgrade in D and a full TC to teach the system will help.

      They seem to be getting better at is as well, but this year is pretty much lost.

    6. woodguy
      March 30, 2013 at

      I know that 98-83 were not getting many 2nd shots in the ozone as per your post on it, but I still think that the increase in SA is heavily related to the system which they are not good at yet.

    7. Oilerfan-atic
      March 30, 2013 at

      Tyler I always like your analysis…when I think of why the Oilers are not playoff bound I think: Jones, Petrell, Khabby, Brown, Potter, Peckham, Whitney, Fistric & Belanger….not many questions to be answered when 40% of your lineup is either washed up or never was NHL calibre….to have these slugs in your lineup after being permanent cellar dwellers boggles the mind.

      • Laura
        May 31, 2013 at

        The problem with Linus Omark is that he is ahotner small player and Edmonton is stock full of skilled small players. They’ve been pushed around for the last 2 years due to their size. They need the size to make the room for these small player to play. The number one thing the Oilers need to due is dump that bum Horcoff. I don’t care how they do it, but just punt his sorry butt off the team, then draft Seguin to replace him.

    8. Showerhead
      March 31, 2013 at

      Not trying to oversimplify, but couldn’t this data just be a compelling argument regarding the difference between having Penner or Horcoff or Smyth on your line and having an 18 year old rookie?

      If you replaced today’s Nail Yakupov with a 2006 version of Ryan Smyth, for example, would it be reasonable to suggest:

      1) The SA would decrease
      2) Smyth’s shot totals would decrease the overall % of Gagner/Hemsky shots
      3) Puck recovery and overall possession numbers would improve?

      Basically I’m taking Ryan V’s list and looking for an alternate plausible explanation.

    9. PaperDesigner
      April 1, 2013 at

      I am curious, Tyler. Has there been any significant movement in the numbers since Paajarvi was put on that line? Small sample size, I understand, but I wonder if the context of who Gagner and Hemsky have been on the ice with might shed some light.

    Leave a Reply

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