• Same Old

    by Tyler Dellow • November 18, 2013 • Hockey • 7 Comments

    Early in the season, I was pretty optimistic because the Oilers were hanging around with the opposition and, if not for some spotty goaltending, would have gotten some results. Their Corsi% when the score was close was really good – through the game with Pittsburgh, they were putting up a 48.4% Corsi% in that situation. (Score close means that the game’s within a goal in the first two periods or tied in the third.)

    If Dallas Eakins’ team had simply accomplished that over the course of a full season, I’d be ready to call his first season a success. It would have been the best Corsi% in close situations by an Oiler team in the seven years for which we have this data and a monstrous step forward from last year’s disastrous 43.4% Corsi% in these situations. If you tacked five points onto the Corsi%, you could reasonably say that the ship had turned and was heading in the right direction.

    Over the last 15 games though, it’s been worse than last year. The Oilers have a 41.1% Corsi% in close situations starting with the game against the Islanders. This should go without saying but you cannot expect to have any sort of success doing that. Get outshot 59-41 when the score’s close and the odds are you aren’t going to be on the happy side of things when the score’s no longer close.

    One of the popular theories as to what’s wrong with the Oilers relates to their having played a lot of games against the Eastern Conference. I don’t really buy it. I went through and created a weighted Corsi% for the Oilers’ opponents in the first seven games and the last fifteen games. I weighted each opponent’s Corsi% close based on the percentage of Corsi% close events that occurred in that game. For example, Winnipeg’s got a 48.1% Corsi% close right now. There were 61 Corsi events when the Oilers-Jets game was close. There were 351 Corsi events in the first seven games with the score close. That’s 17.37% of the Corsi events in that period. 17.37% * 48.1% = .0835. Repeat etc.

    If you do this, you find that the weighted Corsi% close of the Oilers opponents during their first seven games was 48.9%. In the next fifteen games, it works out to 50.6%. The opposition got about two points tougher. The Corsi% in close situations dropped 7.3 points. This doesn’t really make sense to me.

    If you look at it on an individual basis, it gets weirder:

    There were a lot of really good signs from individual players early in the year. Hall was really dominant after his disastrous two games at centre – the Oilers got out-Corsid with the game close 24-10 with Hall at centre. When he moved to the wing, they went 56/44 with him on the ice and the score close in the next five games. Since then – and I recognize he’s coming off an injury – he’s Lennart Petrell, who posted a 38% Corsi% with the score close during 2011-13. This is a problem.

    There’ve been whispers of changes in how the Oilers approach things from a tactical perspective and I wonder if this links back to that. Dallas Eakins mentioned during a media availability that they’d changed things in the defensive zone and abandoned the much discussed swarm. The time period that he suggested that the change was made roughly corresponds to the point between the Pittsburgh game and the Islanders game. I’ve felt, watching the games, like we’re seeing the Oilers forecheck much less aggressively and basically let the opposition come to them, in addition to there seeming to be little beyond gunning the puck out of their own end when they do get control of it.

    I don’t trust my eyes enough to say for sure that this is the case – I have to sit down and re-watch games in order to get comfortable with that, contrasting what was happening earlier with what’s happening now in order to get comfortable being sure about that. That being said, it sure does seem like something was changed and, whatever it was and whatever Eakins and Craig MacTavish say, it’s hard to get the sense watching the Oilers that they’re getting closer to a breakthrough. The numbers say they’re going backwards.

    Email Tyler Dellow at mc79hockey@gmail.com

    About Tyler Dellow

    7 Responses to Same Old

    1. Jay
      November 18, 2013 at

      Actually watching them, they now look every bit as Lars Eller described. Maybe I’m wrong but there seems to be no structure to their attack? They are getting outclassed/outcoached every game now.

    2. godot10
      November 18, 2013 at

      Fooled by Randomness?

      CC CC avg CC SD CC SE
      1 49.2% 0.4920 #NUM! #NUM!
      2 28.6% 0.3890 0.1457 0.1457
      3 46.2% 0.4133 0.1113 0.0787
      4 53.8% 0.4445 0.1102 0.0636
      5 50.8% 0.4572 0.0996 0.0498
      6 51% 0.4660 0.0916 0.0410
      7 47.1% 0.4667 0.0837 0.0342
      8 41.5% 0.4603 0.0796 0.0301
      9 25% 0.4369 0.1022 0.0361
      10 38.2% 0.4314 0.0979 0.0326
      11 51% 0.4385 0.0959 0.0303
      12 54.5% 0.4474 0.0965 0.0291
      13 26.3% 0.4332 0.1056 0.0305
      14 57.9% 0.4436 0.1087 0.0301
      15 33.3% 0.4363 0.1085 0.0290
      16 40.7% 0.4344 0.1051 0.0271
      17 46.2% 0.4361 0.1020 0.0255
      18 52.8% 0.4412 0.1013 0.0246
      19 32.4% 0.4350 0.1020 0.0241
      20 35.3% 0.4309 0.1010 0.0232
      21 50% 0.4342 0.0996 0.0223
      22 34.7% 0.4302 0.0989 0.0216

      I took the value of Corsi close of each game as a sample and then did running simple stats. Average, standard deviation in the mean, and standard error in the mean. (Hopefully Google doc formulas are correct).

      After 22 games, the average of the Corsi Close where each game is considered an individual sample is 43.0%. The standard deviation in the mean is stabilizing just under 10%. And the standard error in the mean is nearing two percent.

      But from first game where one can make a statistical estimate of the error bar, i.e. after the second game, 43% falls between the average Corsi Corsi close plus/minus the standard error.

      i.e. All the data from the 2nd game onwards is consistent with a Corsi Close average of around 43% and a width of the distribution having a standard deviation of around 10%.

      The alleged early improvement in Corsi, and the alleged decline in Corsi is just being fooled by randomness.

      • godot10
        November 18, 2013 at

        I forgot to credit extraskater.com for the Corsi data.

        To repeat the point:

        The Oilers Corsi close results are consistent with a normal Corsi Close distribution with a mean of 43% and a standard deviation of just under 10%.

        No early improvement. No subsequent falloff.

    3. Marek Tyler
      November 18, 2013 at

      Why was Smid traded? Yes, his corsi is borscht but he’s an actual defence man. Was it for cap space? Was it for the goalie prospect? Did he secretly like Knickelback? Are they better now

      Statistically speaking (age, games played, lunar cycle) when will RNH start winning over 50% of the draws?

      Is a “Pronger” or “Weber” the magic bullet or can the defence and team be successful in the manner they are currently build and coached?

    4. Destroyko
      November 18, 2013 at

      I’m not sure how the timelines sync up, but I recall Eakins saying that some guys didn’t know what standard D coverage is. If that’s the case, it would make sense to revert to teaching that basic coverage, and then re-instituting the swarm on top of it. You can’t teach calculus without knowing addition. Hopefully once he’s convinced that guys have the basics down, he can go back to what seems to have been working.

    5. Johnny
      November 18, 2013 at

      Watched the game live on Saturday. The team was so badly outplayed in the first 40 that it was embarrassing.

      Line combos were strange (as always).
      Gagner taking (and predictably losing) an O-zone face-off after 2 consecutive Flames icings was unfathomable.
      No forecheck of any quality.
      Big trouble coming out of their own end — couldn’t get the puck past the red line with possession.
      Horrible puck management in the D-zone that led to numerous high quality scoring chances against.
      Dubnyk (and a little luck) was the only reason the Oilers won that game.
      Yakupov may be a shooter, but he’s a disaster with and without the puck.

      With so much focus on the defensive system, it seems that the team has forgotten how to enter the O-zone and how to forecheck. Too much thinking happening on the ice. Just play the game the way you know how!

      Hall’s breakout down the wing on the first goal was the first time that happened in the whole game. Very unstructured play.

    6. donair poutine
      November 18, 2013 at

      I’m no big storylines guy but I do think that at some point you just give up. When you’re this far out of the playoffs for a team that’s only ever been out of the playoffs, what’s your motivation to go all out? Another tank job’s all in a days work, right?

      By the way, haven’t mentioned it yet but good to see you back posting Tyler.

    Leave a Reply

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