• An Empirical Explanation of Zonestart Effects, Colton Orr is Awful and Greg Cronin Makes No Sense

    by  • October 4, 2013 • Hockey • 25 Comments

    I’ve alluded to the concept of open play Corsi% over the course of the summer, although I’ve never precisely quantified it, which is something that I should do. One of the confounding factors with Corsi% is the fact that players take faceoffs at different spots on the ice and some players take more faceoffs in the offensive zone; other players take more faceoffs in the defensive zone. One of the questions that people like me wonder about is how this affects things.

    Over the summer, I came up with an answer to this question. The way I did it was pretty straightforward. I asked myself this question: “For how long after a faceoff is won or lost is there an impact on play?” There are two types of impact to worry about: the likelihood of the team winning or losing the draw getting or allowing shot attempts and the impact on the volume of shot attempts.

    Hockey allows for tracking six types of faceoff: OZ wins, OZ losses, NZ wins, NZ losses, DZ wins and DZ losses. What I did was look at the 5v5 shot attempts 1 second after the draw, 2 seconds after the draw, etc. for the years 2007-13. Simple enough. We’ll start with offensive zone faceoffs.

    Screen Shot 2013-10-04 at 5.14.27 PM

    What you’re seeing here are line graphs of the Corsi% of a team winning/losing an offensive zone faceoff n seconds after the puck was dropped. So, for example, teams winning an offensive zone faceoff have had a Corsi% of 69.8% in the thirteenth second post-faceoff win over the past six seasons. That drifts down and it isn’t until the 38th second post-faceoff win that they have a Corsi% that’s below 50%. It’s fun to be on the ice when you win an offensive zone faceoff.

    Teams losing offensive zone faceoffs actually have the better of the Corsi% fight for the next ten seconds. When you think about it, this is obvious: if you lose an offensive zone faceoff, the other team still has to get the puck down into your end to generate a shot attempt, an activity that is fraught with risk and peril. From the 11 second mark through to the 18 second mark, they’re then below 50% in terms of Corsi%.

    It’s not just the overall Corsi% that matters here – it’s also the total volume of shot attempts. If the Corsi% gets to 50% quickly, but the shot attempt volume is suppressed, then there’s still a faceoff effect. Let’s look at the shot attempt volumes, both for and against, for each faceoff type. Keep in mind – an OZ+ for one team is another team’s DZ-.

    Screen Shot 2013-10-04 at 5.14.40 PM

    I’ve presented this on a per 100 faceoff basis. We see about what we’d expect. There’s a huge shot volume when a team wins an offensive zone faceoff – the team has possession and are close to the opposition net. It takes until the 23rd second following a OZ-/DZ+ situation until the shot volume equalizes though. Even though the team losing the draw gets to 50% Corsi-wise pretty quickly, there’s still an effect because of the reduced shot volume. I’ve been using 21 seconds as my cutline as far as that effect, along with 37 seconds post-OZ faceoff win.

    Screen Shot 2013-10-04 at 5.14.55 PM

    For neutral zone faceoffs, there’s an effect that lasts for 29 seconds – in the 30th second, the team losing the faceoff is more likely to generate a shot attempt than the team that won it. We know from the graph above that the shot volume impact is gone before that so it’s pretty easy to see that there’s a neutral zone effect that lasts for 29 seconds.

    Screen Shot 2013-10-04 at 5.15.05 PM

    The mathematically astute amongst you will recognize that the chart for defensive zone faceoffs looks a lot like the chart for offensive zone faceoffs. This is because, of course, one man’s DZ win is another’s OZ loss. So the same logic set out above applies and a defensive zone faceoff win sees an impact for ~21 seconds and a defensive zone faceoff loss results in a 37 second disadvantage.

    If you want to talk about open play hockey then, when there are no faceoff effects present, the easiest thing in the world to do is just to knock out the shot attempts within x seconds of a certain type of faceoff. This will, I hope, put an end to the debate about how long the faceoff effect lasts for – this is the empirical answer, I think. The Corsi% in each of these situations is pretty damn consistent, year after year.

    Screen Shot 2013-10-04 at 5.11.24 PM

    (The title is screwed up this graph – it should be “Post OZ Faceoff Loss.”)

    Screen Shot 2013-10-04 at 5.08.16 PM

    Screen Shot 2013-10-04 at 5.08.39 PM

    Screen Shot 2013-10-04 at 5.09.13 PM

    Screen Shot 2013-10-04 at 5.10.34 PM

    Screen Shot 2013-10-04 at 5.09.51 PM

    Is this useful information? I think it is. I was struck by this comment about Colton Orr earlier this summer:

    I think Orr has proven he’s more than just an enforcer. He was used a little on the third line this season, he is one of the more reliable guys at chipping pucks out, knowing the conditions of the game, chipping pucks out and chipping pucks in, changing smartly, and he’s very responsible on the back check.

    This came from a man who is paid money to coach a professional hockey team. A proposition: if Colton Orr is “one of the more reliable guys at chipping pucks out” it will show up in his numbers following defensive zone faceoffs and they won’t be horrible. Reality: the Leafs had a 10.7% Corsi% in the 37 seconds following DZ faceoff losses last year with Colton Orr on the ice. Whether he’s good at chipping pucks out or not, teams are crucifying the Leafs in a situation in which a player’s defensive ability would appear.

    Orr was on the ice for 12% of the Maple Leafs defensive zone faceoff losses at 5v5 in 2013. Amongst guys who were on for at least 10% of their team’s DZ losses, Orr’s 10.7% Corsi% ranks 3,428th out of 3,447 people with enough TOI to qualify. The Leafs were a garbage fire of historic proportions when they lost a DZ draw with Orr on the ice. That said, Orr actually surpassed the figures posted by a couple of other heavies last year: Chris Thorburn, Ryan Reaves and BJ Crombeen all posted even worse figures.

    As far as Cronin goes, you wonder. Is he insane? Is he lying? Is he aware of how poor the Leafs results were in a defensive situation like this with Orr out there? I have no idea. We’re starting be able to measure stuff finely enough that we can ask these questions though. It’s hard to imagine any sort of a defence for the guy as a hockey player, no matter what the Leafs might say.

    Email Tyler Dellow at tyler@mc79hockey.com


    25 Responses to An Empirical Explanation of Zonestart Effects, Colton Orr is Awful and Greg Cronin Makes No Sense

    1. M Parkatti
      October 4, 2013 at

      The area between the blue and red lines suggests that winning faceoffs may be more important than I (& others?) thought. A centre winning draws at less than 40% would really be putting a drag on his linemates…

      • Triumph
        October 4, 2013 at

        As I noted on Twitter, while it’s worth a lot of Corsis, I believe shooting percentage off faceoffs is considerably worse than shooting percentage during regular 5v5 play. This is where Corsi as a proxy for possession is indicative of who won the faceoff, but I’m not sure it says as much as you think about who’s going to be scoring goals.

        I forget where I read that stat about shooting percentage off faceoffs, maybe if I throw up the Eric T. beacon he’ll remember.

    2. Tyler Dellow
      October 4, 2013 at

      Yeah, I think that’s right Michael. That being said, not a lot of guys win faceoffs that infrequently.

      • M Parkatti
        October 4, 2013 at

        I guess I had someone like Cogliano in mind, right around that 40% for his career. With this analysis in mind, you really could make a case that a guy like Cogliano should be played on the wing if his faceoffs skills are so terrible. If a guy’s taking 1000 draws in a year, you’re giving up 100 FO wins using Cogliano over a year. You could calculate what the price of that is in terms of shot differential, and then goals. Might be trivial, but at least this provides a frame of reference for doing the calculation…

        • John Speranza
          October 4, 2013 at

          I’ve always wondered sort of the converse, which is why there aren’t wingers who are faceoff specialists — maybe they’re not strong enough defensively to match up against the opposing center, so a “real” center (a Cogliano type) takes that assignment.

          • October 5, 2013 at

            David Steckel “played” wing more often than not while still taking faceoffs when he was with the Leafs if I recall correctly.

    3. October 4, 2013 at

      The problem with this type of analysis is that it doesn’t take into account the skill level of the players. We know the skill level of players who start in the offensive zone is often quite different than the skill level of those that start in the defensive zone. Without taking into consideration the skill level we don’t know whether the effects are due to the face off location or due to the talent level of the players. You may not be seeing the effect of an offensive zone face off fully gone until 38 seconds because it isn’t until 38 seconds that the lines have changed. When people first started looking at zone start effects they just did a straight correlation between zone starts and corsi% and over estimated the impact because they didn’t account for the fact that players with a heavy defensive zone start generally aren’t as good at hockey (generating offense in particular).

      If I have time I’ll do some more digging into the numbers but I think taking skill level of the players into consideration is important.

      • Tyler Dellow
        October 4, 2013 at

        *buzzer sound* Wrong!

        • October 4, 2013 at

          So publish your data for every player and lets compare it to my data and see where the differences lie. Lets see if what you outline above makes a practical impact on the data. You keep telling me that I am wrong but every piece of ‘open play’ data you have published in articles has a .96 correlation with my data which is about as good as you will get using almost any kind of sampling method. So, are you willing to put your data to the test?

        • May 6, 2014 at

          End of 2 periods agaisnt Anaheim, and Bolland, Shaw and Kruger have all had their share of problems agaisnt ANA I have watched 3 games versus them, and ANA is a better hockey team yes we are without Sharp and Hoss, but not in the first game.Kruger and Shaw are showing their age tonight, and Bolland is a mess. Bowman has done lots to help the Hawks but he is also directly responsible for this huge void at Centre the Hawks have a chance to put ANA away for first in the conference, and we are not showing the man power to do so.

    4. Tyler Dellow
      October 4, 2013 at

      I’m not sure that I ever said that your method didn’t come close. I’ve doubted your rationale throughout for why it worked throughout because it seemed to me that faceoffs had a big impact. You’ve been adamant that faceoffs aren’t a big deal – this shows pretty clearly that they are and that the impact lasts for a while. You’ve been sure that it’s the players taking OZ draws versus the players taking DZ draws – do you really think that the difference is 30 points of Corsi between those players? Get serious.

      • October 4, 2013 at

        “You’ve been adamant that faceoffs aren’t a big deal – this shows pretty clearly that they are and that the impact lasts for a while.”

        Yes, we can look at individual face offs and see that they have an impact but if that impact is dwarfed by all the other ice time a player gets they largely become irrelevant. Show me that they have a big impact on a players statistics because nothing I have seen shows they are, especially beyond 10 seconds. Run a correlation between your ‘open play’ numbers and your straight 5v5 numbers and let’s see what the difference actually is. What you have presented above is a theory that zone starts could have a significant impact. The next step is to show that they actually do and the extent that they do.

    5. Tyler Dellow
      October 4, 2013 at

      FWIW, Dave, I’m pretty sure that the reason that your adjustment works is because what we call “ZoneStart” isn’t really that. It doesn’t measure where you actually *start* your shifts because we sweep in all of the faceoffs that occur during your shift. So in many cases, if a guy’s on the ice for a faceoff in the DZ or OZ, it’s because his line just generated a shot there.

      • October 4, 2013 at

        That may be the case but if it is, shouldn’t the player get credit for having the talent to end his shift in the offensive zone and thus start his next one there too. Why should we eliminate this benefit from the player who earned it?

        You could be right that the actual impact of a zone start lasts longer than 10 seconds, but if eliminating the time beyond 10 seconds has no practical impact on a players stats, those extra seconds are essentially irrelevant. Of course, if the benefits of those those extra seconds are the equivalent of those zone starts that the player himself ‘earned’ by ending his previous shift in the offensive zone with a shot on goal then for player evaluation purposes it is better that we keep that data, not toss it aside.

    6. Tyler Dellow
      October 4, 2013 at

      Remember when you were nagging me to publish this data and I sarcastically pointed out that whatever it showed, you’d claim you were right? This isn’t a theory: faceoffs have a massive impact on what follows. Your rationale has been debunked. What do we get? This.

      • October 4, 2013 at

        You have yet to show me that zone starts have “a massive impact” on a players statistics. Do a comparison of “open play” stats vs “5v5 stats”. Which players have the largest impact and what is the magnitude of that impact. Prove me wrong with actual examples of ‘massive impact’ on a players stats.

        • chuck
          October 5, 2013 at

          Example – Grabovski?

          You can compare his usage over the last 3 years and see his production. You’ll get more data when he’s again placed in more OZ starts

    7. Tyler Dellow
      October 4, 2013 at

      That may be the case but if it is, shouldn’t the player get credit for having the talent to end his shift in the offensive zone and thus start his next one there too. Why should we eliminate this benefit from the player who earned it?

      I dont have a problem with crediting the guy for that but it doesn’t really get into the faceoff effect. In addition, it kind of pollutes the numbers for doing a team by team comparison.

      All of this is irrelevant. You don’t get to take explanations that come up when someone actually examines all the data and figures out what’s going on and say “See I was right all along even though I had an entirely different rationale that didn’t pass the smell test?”

      • October 4, 2013 at

        We are coming at this from different angles.

        You: When we look at zone starts, they seem to have dramatic effects on what happens in the following ice time.

        Me: Those differences don’t have a significant impact on a players statistics so for player evaluation purposes there is no need to make an adjustment (beyond 10 seconds).

        We can both be right. I just don’t know what you call me wrong.

        This is equivalent of:

        “Playing against Sidney Crosby has a significant impact on your corsi as compared to when you are playing against Colton Orr”


        “Nobody plays against Sidney Crosby so much that he has a major impact on their stats over the long-term”

    8. Donair Poutine
      October 4, 2013 at

      Are there S% effects from a face off win as well? And if so, how long do they last?

    9. Pingback: Why I have become more open-minded about hockey analytics — and a blurb on Morgan Rielly | Toronto Maple Leafs

    10. Pingback: Boys on the Bus » The Influence of Face-offs and Zone Starts on Corsi

    11. Pingback: Boys on the Bus » The Impact of Winning Faceoffs in Different Zones

    12. Pingback: On-ice shot distance | Extra Skater Blog

    13. June 30, 2014 at

      Because all my life I was sexually torn because of
      emotional abuse and emotions of guilt. The long phrase effects of many of
      these sexually transmitted disease can last for a

    Leave a Reply

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