• Big Oilers Data IX: Neutral Zone Faceoff Wins

    by Tyler Dellow • May 21, 2013 • Hockey • 5 Comments

    This is part of a series looking for reasons for the Oilers Corsi% collapse in 2012-13 by examining things on a shift-by-shift basis. Part 1 can be found here. Part 2 can be found here. Part 3 can be found here. Part 4 can be found here. Part 5 can be found here. Part 6 can be found here. Part 7 can be found here. Part 8 can be found here.

    As you might have guessed from the last few weeks of posts, I’m pretty intrigued by the information that we can tease out of the data when we break hockey games down on a shift-by-shift basis like I’ve been doing with the past two years of Oiler games. I’ve been taking a very micro approach to the data so far; in doing so, I think I might have skipped a bit of an intermediate step that I want to remedy today.

    One of the difficulties in looking at data like this is that it’s so new that we don’t have any benchmarks for what’s good and what’s bad in a specific situation – we’ve got data from one (bad) team for two seasons. If other people see value in what I’m doing and this becomes a thing, a way that we deal with the data from hockey games, in time we’ll learn what the benchmarks are. It’s like anything else that’s data and new – when you first dip into it, you’ve got to spend a fair amount of time figuring out what it means.

    As those of you who have been reading this series as I’ve gone along will be aware, I’ve been kind of looking at things on the basis of eight different kinds of 5v5 shift: Open Play (no faceoff during shift), six types of shift with one faceoff (OZ+, OZ-, NZ+, NZ-, DZ+, DZ-) and multi-faceoff shifts. The cool thing with seven of those types of shift is that I can get a benchmark of a type by looking at how the Oilers opposition did in the same situation.

    For example, I can compare the Oilers’ Corsi% performance on OZ+ shifts to how the Oilers opponents did in OZ+ situations simply by subtracting the Oilers’ Corsi% in DZ- situations from 1. In 2012-13, the Oilers had 27.7% of the Corsi events in DZ- shifts; 1-.277=0.723; the Oilers’ opponents had 72.3% of the Corsi events in OZ+ situations. It’s a kind of useful benchmark, although we need to be aware that the Oilers were a below average team and we’d expect that other teams would do slightly better against them than whatever the true league average benchmark might be.

    In order to take a look at this, I calculated numbers on the basis of the average performance of eleven forwards in seven of the eight situations that I’ve identified. The eleven forwards I’ve included are Eric Belanger, Jordan Eberle, Sam Gagner, Taylor Hall, Ales Hemsky, Shawn Horcoff, Ryan Jones, Ryan Nugent-Hopkins, Magnus Paajarvi, Lennart Petrell and Ryan Smyth. These are the eleven forwards who played significant roles on both the 2011-12 and 2012-13 Oilers.

    I simply took an average of the Corsi% for these eleven guys in each of the seven situations that I’ve outlined and I’m treating that as the Oilers number. You can see at the left what the data shows. Big picture observation: the Oilers did worse in every aspect of the game relative to the opposition in 2012-13 than they did in 2011-12.

    Let’s talk about 2011-12 first. The good news, I suppose is that the Oilers were as good as the opposition in 2011-12 on OZ+ and DZ- shifts (two sides of the same coin, obviously). They were marginally worse on open play shifts. They did 5.6 points worse than the opposition did with OZ- and DZ+ shifts. The biggest difference between the Oilers and their opponents was in NZ+ and NZ- shifts; on both of those, they were 7.6 points worse than the opposition.

    It strikes me as interesting that this season lines up pretty much exactly the same way. The closest that the Oilers came to their opponents was during open play shifts – a 7.7 point difference. Then, we see OZ+ and DZ- shifts, on which the Oilers were 9.5 points worse than their opponents. Next, DZ+ and OZ- shifts, at 10.3 points worse than the opposition. Finally, we see NZ+ and NZ- shifts. On both of those, the Oilers performed a whopping 19.6 points worse than the opposition.

    An observation about NZ shifts. If you look at the Oilers’ NZ- numbers in 2011-12 and 2012-13, you can see that they’re awfully similar: 41.5% in 2011-12 and 39.3% in 2012-13. When the Oilers played shifts on which they lost a neutral zone faceoff in 2011-12 and 2012-13, they were awfully close being the same, year-over-year. The reason for the collapse, relative to the opposition, in NZ situations lies in what the Oilers did on NZ+ shifts.

    They weren’t particularly good at NZ+ situations in 2011-12 but in 2012-13, holy cow. They were worse in NZ+ situations in 2012-13 than they were in NZ- situations in 2011-12. It’s astonishing. Can we tease a little bit more out of the data? I think that we can. In 2011-12, the Oilers generated at least one SAF (shot attempt for) on 37.6% of their shifts. Last year, they generated at least one SAF on 34.7% of their shifts. So a bit of a change, but not a huge one.

    What about multi-SAF shifts, the percentage of those shifts on which they recorded an SAF that turned into multiple SAF? Well, in 2011-12, 30.1% of their shifts were multi-SAF shifts. In 2012-13, that number was 18.4%. So, um, there’s a 40% drop or so, which seems sort of significant.

    Corsi% is not built on SAF alone but also on the volume of SAA (shot attempts against). On NZ+ shifts in 2011-12, the Oilers allowed at least one SAA 36.1% of the time. This year, that number ballooned to 43.6%. Curiously, the percentage of multi-SAA shifts actually dipped slightly: from 29.1% to 27.4%.

    It strikes me as awfully interesting that the three of the four multi-SAF/SAA percentages are so similar: 30.1%, 29.1% and 27.4%. We’re basically measuring the same thing here, the ability of the team that’s generated at least one shot attempt in a given situation to turn it into multiple shot attempts. That Oilers number this year, 18.4%, stands out to me like a beacon in a storm when I’m looking for an explanation as to what went wrong with the Oilers. If you’re generating more multi-SAF shifts, it means that by extension, the puck is in the other team’s end and they aren’t coming to your end.

    Why might this have occurred? The explanation that seems most likely to me is that the Oilers were doing something differently this year that impacted primarily on their ability to recover pucks after shots were taken and permitted the opposition to recover the puck, exit the zone and head towards the Oilers end of the rink. Let’s ask the classic Bill James question here: If that were true, what might we expect the data to show? I’d think that the data might show that the Oilers had a greater percentage of NZ+ shifts this season on which both teams got a shot on goal.

    Does the data show that? It does, although truthfully, the bigger problem is the increase in the volume of NZ+ shifts on which the Oilers generated 0 SAF and allowed 1+ SAA.

    I want to pause and make a point about something. Certainly, this isn’t an issue that the mainstream media or the Oilogosphere was aware of. In the millions and millions words spilled about the Oilers this season, I don’t recall anyone saying “They don’t do enough on shifts where they win a neutral zone faceoff.” It’s plainly obvious when you look at the data like this but nobody was discussing this during the season. Were the Oilers aware of this specific problem? It’s a great question, albeit one without a readily available answer.

    Inevitably, this leads to a question of “Why?” Why were the Oilers so poor at generating SAF on NZ+ shifts last year relative to what their competitors did in the same situation? Why did they get so much worse this year? A smart technical hockey mind could probably address this by reviewing some more video, once he was aware of the problem. A less trained hockey mind but generally bright dude *cough* could probably come up with some ideas if he looked at video, although it’s a daunting task.

    When touch data, a record of when and where every touch on the ice takes place, comes into being – and it will, and I’d be somewhat surprised if there aren’t teams doing it already – a team could do some pretty amazing things in terms of identifying where things are breaking down for them versus the rest of the league. As a tool for isolating a specific issue like this, it would be wildly useful. Given that what we’re seeing here runs through all of the eleven forwards I’ve identified, who can be found at the top and bottom of the roster, I’m inclined to think that there was a tactical change that really didn’t work this year. It just seems like the most logical thing to me, although I don’t think that’s proven by this data.

    There’s a fair amount of content to chew over here, so I think I’ll cut this one off here and come back to the other things that the data suggests in the next post. If you’ve just scrolled down to the end of this post looking for conclusions, they are as follows: 1) the Oilers regressed in each of OP, OZ+, OZ-, NZ+, NZ-, DZ+ and DZ- shifts this year, 2) they were particularly poor at generating Corsi% on NZ+ shifts in 2011-12 and horrifically so in 2012-13 and 3) a significant part of their inability to generate Corsi% on NZ+ shifts is tied to an inability to generate multi-SAF shifts.

    Email Tyler Dellow at mc79hockey@gmail.com

    About Tyler Dellow

    5 Responses to Big Oilers Data IX: Neutral Zone Faceoff Wins

    1. Woodguy
      May 22, 2013 at

      I think this is the point where zone entry data can help point in finger of blame.

      Maybe not at players in particular, but at the system or break downs therein.

      Would help to have last year’s zone entry data too.

      I would help gather that data but *looks at watch and taps it* I got….a thing…

      Also,

      Are you already data mining this info for the rest of the teams and dumping them into a database?

    2. Tyler Dellow
      May 22, 2013 at

      Yeah, Darcy, I’m thinking the same thing. It’d be fascinating to be able to cross-reference this with that information. I should drop Willis a note – maybe we can do a Dellow/Willis co-production.

      Not nearly at the point of doing this with other teams. Will get there eventually.

      • Woodguy
        May 23, 2013 at

        What I’m most looking forward to is teasing your exactly why Hall will drive his line mates over 50% while the other flounder.

        I’ve suspected its his ability to carry the puck from blue to blue and maintain possession, and often credited 93 with some of it by his ability to get the puck to 4 in the right spot with the breakout pass.

        Is 4 overcoming the system or is he the only one who can play it correctly

        Mind you, if he’s the only one playing it correctly on a team of (mostly) NHL pros, perhaps the system should be changed.

        Also,

        If you plan on going forward with wide scale data crunching in this vein and making it public, I would be interested in helping defray some of the costs. PM me on twitter.

    3. Pierce Cunneen
      May 23, 2013 at

      Awesome work

    4. Pingback: What is 'open play' hockey? - HockeyAnalysis.com

    Leave a Reply

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