• Eakins on Defending A Lead

    by  • September 25, 2013 • Hockey • 10 Comments

    “It kills me. It makes me crazy. I was getting ready to check myself in in the third period because we started to sit back and I don’t understand that, why you would, after having great success for two periods by pushing the pace, and limiting their time and space, why would we suddenly give them time and space when the other was working for us? I feel the best way to defend a lead is to score a goal. That’s the way to do it. To sit back and give them chance after chance doesn’t make sense to me…”

    -Dallas Eakins, on protecting a lead

    I think that there’s a lot of logic to what Eakins says and I think those of us who are interested in tactics and coaching are in for a fun couple of years. That being said, I know that there have been 180 team seasons in the BTN Era (30 teams, six seasons each) and I know that 180 teams have been done worse in terms of their Corsi% when they’re leading than they’ve done when they’re tied.

    Sure, you might say, teams do worse with Corsi% when they have the lead but they’re giving up lower quality shots, so it’s a good tactic. Here’s the thing with that though: teams that are leading games have been outscored 7937-7630 during the BTN era. This is all the weirder when you consider that a good team is more likely to be leading a game than a crappy one. Do they enjoy an edge in the percentages? Sure. 8.7% shooting percentage and a .924 save percentage. It doesn’t make up for the overwhelming edge in shots that they give up though. Only 13/180 teams have managed to post a Corsi% above 50% while playing with the lead since 2007-08.

    It takes two teams to move the Corsi% needle but my sense is that teams playing with a lead are probably too conservative. (Oddly, the team that saw the lowest decline when playing with a lead was the 2013 Oilers, whose 42.5% Corsi% with the lead was only .3 percentage points than their Corsi% when tied. Krueger was asked about whether the team sat back with a lead and he said he told them to do exactly the same thing. So it would appear.) I haven’t done anything exhaustive on it – I’d want to look into whether there are some teams that are able to consistently make the percentages work for them or something but the bottom line, to me, is this: 108 of the 180 team seasons since 2007-08 saw teams get a higher percentage of the 5v5 goals when they were tied than they did when they were playing with the lead.

    What might make coaches conservative? I suspect that coaches are willing to live with goals that they see as being part of the randomness of hockey – think about Yakupov’s goal against Los Angeles last year. A shot from the point is blocked, Taylor Hall whacks it towards the net, it pops in the air and Yakupov knocks it out and in. Bounces and some skill. It’s easier to kind of mentally write that off as bad luck than it is if you give up a two on one that leads to the tying goal and you know that you could have prevented that odd man rush if you’d told your players to keep four men above the puck at all times. What you maybe don’t see is that doing that, or something like it, makes the time in your own end more likely and the bad luck goal that pinballs in more likely.

    In any event, this will be one of many interesting things to follow with the new coach. He seems to have articulated a view as to how things should be done that’s at odds with what his fellow coaches but that is also plausibly defensible, at least on a cursory glance. It’s worth adding to the list of things to keep an eye on with this team.

    Email Tyler Dellow at tyler@mc79hockey.com


    10 Responses to Eakins on Defending A Lead

    1. daryl
      September 25, 2013 at

      This one feels like a hornet’s nest of variables. You touched briefly on the “2 teams” idea, but its worth looking into the other team more closely maybe also – was the leading team’s corsi lower because they were more conservative, or is that the result of the trailing team becoming more wide open?

      Then you get to the next level of analysis on what is the actual cause of this – shortened benches when you have a lead? What are the average numbers for the players who actually play more when your team has a lead? Let’s say you roll 4 lines and 6 D and then get a lead and drop that to 3 lines with your topC double shifting and go down to your top 4D. I guess you could argue that’s the cause of a lower corsi and an indication of a coach being more conservative, but it isn’t an indication in a difference in style of play necessarily.

      My brain hurts on this one.

    2. woodguy
      September 25, 2013 at

      I’ve bever understood “prevent” defence in Football or Hockey.

      Stop doing what gave you the lead and let the other team come at you.

      In hockey the most common reasoning I’ve heard is to avoid “giving up odd man rushes”

      So they go from 2 fore checkers deep to one and try to plug up the neutral zone.

      They also start to focus on “just clearing the zone” instead of exiting the dzone with the puck and willfully give up possession.

      I guess with the high SV% of the average NHL goalie will generally ensure that the increase in shots against won’t result in more than one goal, but you’re still playing with fire.

      If the most dangerous lead in hockey is two goals, then you better make it three.


      Eakins post game presses are full of awesome.

      He was asked to comment about “Perron’s hands” after he had two sweet tips for goals in the game.

      His answer: “He has one left one and one right one”

      My man crush continues to grow.

    3. Waterloo
      September 25, 2013 at

      If one looks at football, it is pretty obvious that teams play too conservatively under “normal” conditions. The result is that even when teams don’t play prevent D, offenses move the ball better near the end of halves and the game because they are forced to adpot a more aggresive (and hence, more optimal) strategy. When the defense piles on with a prevent D, this effect is merely magnified.

      If we move this to hockey, perhaps the “default” way that NHL teams play is too conservative, and that, by moving to more aggressive strategies when behind, teams are actually playing more optimally, regardless of what the opposition does. If this is the case, then teams that try to play the same when ahead as they do when tied will have worse results; furthermore, when teams do adopt more passive styles (1 fore checker, off the glass and out, always dump it in, etc), this difference is compounded and the team behind looks like world beaters even if they suck in “normal” conditions.

    4. ironsight
      September 25, 2013 at

      I’ve posed this very issue to a former NHL coach – why would you willingly adjust to a strategy that permits a greater number of shots against simply because you have a lead? The answer – there is no change in tactics by the coach. Rather, he claimed the change on his bench resulted from the change in mindset of the players. Rather than continue to be aggressive on the forecheck and attempt to maintain possession on entry into the offensive zone, there becomes a tendency to play more ‘conservative’, to ease up on the forecheck, and to dump the puck deep.

      This seems like the kind of issue that could be overcome with more assertive coaching, but this former coach was adamant that it was a change in the players while leading, not a change in coaching tactics, that altered the shot differential.

      • September 25, 2013 at

        My dad, who coached younger guys for years, told me something similar. That his players would often naturally sit back with a lead, even if he told them to keep attacking.

        • Quellan
          September 25, 2013 at

          Honestly, this matches a lot of my experiences when playing Ultimate Frisbee. It’s a well known fact that the losing team almost always makes a comeback of sorts in the second half. And often preventing this is a point of emphasis for the team and it happens regardless.

          Although you might be able to make the claim that we’re actually conditioning in such a way that this isn’t unsurprising. something like: Goals are correlated with positive corsi events. If we look at the total history of corsi events we get one percentage. The times that goals for a leading team occur correlate with spikes in that percentage, so when we look at corsi when a team is leading, we are looking at the entire history minus a subset of events that are correlated with spikes in corsi. So it makes sense that the remainder is lower than the average.

          That said, I haven’t studies stats/probability enough to be really comfortable in this analysis. It seems like there’s some truth here, but I highly expect that it’s only a grain of it.

    5. Adam Dyck
      September 25, 2013 at

      Obviously “prevent” defence results in you getting less shots on net. Does it also bring down the other team’s total shots (even if they still end up with a net gain)?

      • Jesse Dahl
        September 25, 2013 at

        Assuming there is some data out there that the number of total shots goes down for both sides (i.e. low event hockey), and I’m pretty sure there is, do you think the following could be true?

        At some point in the game it would be more optimal to just sit back even though in the long run it is a losing strategy. In other words, a goal for is less valuable than a goal against is damaging when you’re near the end of the game. If you can lower the variance over that small period of time you will get 2 points more often (and hopefully?) more points on average.

        To use an extreme example to illustrate the point, with only 30 seconds left (or less) in the game, up by one, prevent defense is obviously optimal over a more aggressive style even though it is more likely you will score a goal with a more aggressive style. A two goal lead is not as valuable here as preventing a goal.

        This may be similar to a gambling/bankroll issue.

    6. David
      September 25, 2013 at

      I think you’re missing the point with this argument.

      The point of playing conservatively is to lessen the incidence of the opponent scoring. To properly evaluate whether the strategy is a good one, or not, you’d have to determine whether playing conservatively lessens the number and quality of shots so that it’s less likely the opposing team will tie the game, even if they have the edge in CORSI.

      But the greater consideration your missing, in my mind, is that, psychologically, a team that is sitting on a multi goal lead will likely have less motivation to play at the same tempo as when the score was tied. With that in mind, playing conservatively will allow them to keep the number of chances at a minimal level while putting in less effort than the team that is trying to tie the game up.

    7. Steven Olson
      September 25, 2013 at

      Although I certainly agree with the motivation argument made by David, and have experienced similar changes in play due to player psychology in the recreational sports I have played I am going to propose something entirely different as the reason for teams universally having a lower CORSI % when playing with the lead then they do tied or behind.

      Corsi events do not happen with regularity and there is an element of randomness to their occurance. We never see a game that is 50% corsi that completely alternates corsi events. Teams always have times were they pile up the corsi events for, and other times when they pile up the corsi events against.

      And given that generally an increase in corsi events causes an increase in goals, a team will be performing above their natural corsi level when they get the lead, and then naturally regressing to the mean. Alternatively, when a team gets behind, they will have generally been performing below their natural corsi level, and regress to the mean, increasing their corsi % when playing from behind.

      I know its more complicated than that, but I think a purely random model of corsi events without considering tactics can completely explain why teams universally have a lower corsi percentage when playing with a lead. Because they performed above their natural level temporarily to get the lead in the first place.

    Leave a Reply

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