• Taylor Hall, Two Games II

    by  • January 21, 2014 • Hockey • 5 Comments

    One of the things that I find odd about discussion that surrounds hockey is how little discussion there is of why teams lose that can’t be fundamentally boiled down to some magical belief. There’s been a lot of talk over the years in Edmonton about how the team needs to be bigger; there’s been precious little explanation from the GRITENSITY set about how a given third or fourth liner being bigger helps Taylor Hall be better or how he makes Jordan Eberle better. It seems like there’s a belief that these bigger players are just somehow expected to make them better because the Oilers’ average size goes up. You know, magic.

    I tend to think that hockey’s a little more sophisticated than that. The idea at the heart of what numbers do is that you can figure out what sorts of players and tactics increase your chances for success. Not only that, but you can prove your ideas. First it was goal difference – showing that goals are random and that we can reasonably think about a hockey team in terms of how much goal difference it creates. Second it was breaking down goals themselves into pieces and learning what the spread of talent is like with shooters. That led to understanding how fundamental the ability to create and prevent shots is. This leads to yet another level – how are shots created? How are they prevented?

    A wiseass might say “We knew all this already.” While it’s completely untrue – someone reading this today who wasn’t around a decade ago would be stunned at how much resistance there was to the idea of goal difference – it also misses the point. If something is true, I (or you), should be able to prove it with something other than an assertion. If we’re just going on assertions, what makes what Guy A says any more reliable than Guy B?

    So it is with the Mystery of Taylor Hall. Lots of people have lots of ideas. A lot of it just sounds like buzzwords to me. It’s like the criticism that Hall’s lousy defensively. There are things that you can say if you’re in the business of talking about hockey that won’t blow up on you. Criticizing the defensive play of offensively skilled players is one of them – it’s like an inverse Nichols’ Law of Catcher Defence.

    I wrote a post yesterday comparing two games that Hall had played against the St. Louis Blues. One, on February 29, 2012, he had a Corsi% of 64.5%. The other, on December 21, 2013, he had a Corsi% of 33.3%. These are extremes but they kind of sum up the issue with Taylor Hall this year: his Corsi% is in the tank. You don’t win hockey games in the long run with star players who have lousy Corsi% and Hall was always pretty good so it’s worrisome.

    I showed yesterday that these two games were basically the same for Taylor Hall in terms of how much time the Oilers spent in each of the ice when he was out there and how the two teams gained the zone. I’ve updated the table detailing how the two teams gained the zone when Hall was on the ice – a couple mislabelled entries snuck into my data – and I’ve put it at left. The first line in each of the three tables to the left refers to the February 29, 2012 game against St. Louis. The second line is the December 21, 2013 game. It looks like a carbon copy.

    Let’s back up a second here. Craig MacTavish, in his lengthy interview with David Staples, mentioned that he thought Hall was doing better this year, data be damned. I buy that to an extent. The example that MacTavish cited, and presumably what he was talking about, was an instance against Tampa Bay in which Hall took what the defence gave him, chipped a puck in and the Oilers ended up scoring. My data doesn’t really address that kind of situation – I look at how the Oilers gain the blue line. It’s worth thinking about whether it should be how they attack the defence.

    On the broader point though, the scoring chances and Corsi%, MacTavish simply cannot be correct. Hall’s been a player throughout his career who out-Corsis and out-chances. He’s gone off the rails this year and it’s a problem that needs to be fixed. Let’s think about reconciling that with my data from the two Blues games. We see that the Oilers performance with him on the ice looks pretty identical in terms of how the zone was gained and how much time was spent in each zone, year over year. But there’s a big time difference in terms of Corsi%.

    I actually take a little bit of comfort from that. If the Oilers are still generating a big edge in carries into the offensive zone with Hall on the ice, I’m all the more comfortable that it’s not a physical issue or something disastrous. They still seem like a very talented team when he’s out there in terms of this. It must be something else. Let’s look at how the Oilers shot attempts with Hall on the ice in these two games against the Blues were created.

    The first line in each table is February 29, 2012 and the second is December 21, 2013. The big difference between the two games is pretty obvious: in the February 29, 2012 game, the Oilers were creating a ton off pucks that were carried in to the zone. They created almost twice as many shot attempts from pucks that were carried in with Hall on the ice in the first game as they did in the second. I may have missed it but I don’t recall anyone mentioning that during the evisceration of the Oilers that followed that broadcast. Everything else, there isn’t a ton of sample, although generally, the February 29, 2012 game was better.

    It’s important to emphasize that we’re comparing an excellent game with a poor one for Hall – this isn’t a comparison of his typical game in each season. Some games, you jump on the ice and the puck’s in the right end and you get in on generating a few shots. Some games your centre wins faceoffs in such a way that you generate shots. That’s the vagaries of Corsi%. In 2011-12, the Oilers generated 58.5 SAF/60 at 5v5 with Hall on the ice; in the game I’m looking at, that number was 76.4 SAF/60. So there’s about five extra SAF in there than we’d see in the typical Hall game from that season. This year, the Oilers generate 47.0 SAF/60 with Hall on the ice. If he’d had an average game in terms of SAF against St. Louis, we’d expect him to have been on the ice for about 9 SAF instead of 7.

    We’ve got our first set of questions though. Are the Oilers creating fewer shots off carries with Hall on the ice this year or was this a one-off? If they are creating fewer shots off carries with Hall on the ice, why? Have they made some change that’s led to this? Up next, we’ll look at some video from the two Blues games and see if we can’t pick something out of them.

    Email Tyler Dellow at tyler@mc79hockey.com


    5 Responses to Taylor Hall, Two Games II

    1. dawgbone
      January 21, 2014 at

      Are teams adjusting to the fact that the top 6 is full of guys who love to carry the puck into the zone?

      I mean shit we do this in men’s league. Send defensive pressure onto the puck carrier. Sometimes it back fires defensively and they get a really good chance out of it, but most of the time you snuff out the play completely.

      • Tyler Dellow
        January 21, 2014 at

        I’m not sure that they are. Time of possession was pretty similar. They’re still getting the puck into the zone. They just don’t seem able to do anything with it when they’re there.

    2. woodguy
      January 21, 2014 at

      If someone is gaining the zone and then does a soft dump to the corner, are you counting that as a carry or a dump?

      If dumps are shooting in outside the blue and any possession inside the blue is a carry, should there be sub-category of soft dump when they gain the line but the knowingly give up possession once inside the zone?

    3. Woodguy
      January 21, 2014 at

      This got me thinking about corsi as a proxy for zone time.

      I know that Vic looked at this back on IOF when the NHL actually published zone time and found a reasonable correlation.

      A WAS blogger also stop-watched a game or two and found close similarity, but that sample isn’t much of a sample.

      If Hall’s corsi can drop 40% or so between two games with only a ~5% drop in zone time, that assumption might need to be re-visited.

      • January 22, 2014 at

        My guess is that kind of extreme variance exists primarily because we’re looking at just two games. We see the same thing with scoring chances and shot attempts. In a single game (or even playoff series) there might be a pretty big gap between them, but that gap narrows significantly over time.

    Leave a Reply

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