Here are the problems with this research.
- There is no baseline. There won’t be until one or more teams decides to challenge orthodoxy and try not pulling their goalie when trailing by one or two goals near the end of the game.
- The study assumes that opposing teams will not change their behavior if teams start pulling their goalie early. The most obvious anomaly will be teams that smartly decide to start taking potshots at the empty net rather than just clearing the puck out of their zone.
- The study assumes that the chances of scoring a tying goal in the last few minutes of a game are the same as they are for the rest of the game (see 1). It is more likely that teams trying to tie a close game will intensify effort as the game draws to a close and generate more scoring chances regardless of whether they have pulled the goalie or not. Teams defending a lead tend to give up opportunities to score (pinching at the blue line, for example) in favor of a tighter defense. Perversely, this allows the trailing team to play more aggressively (eg. pinching at the blue line). (My impression this year is that more and more teams are actually continuing to attack during these periods– which, I think, is smart.)
- The study almost idiotically asserts that you should pull your goalie by the middle of the 2nd period if you are trailing by 4 goals or more. I would love to see a team try this so we can settle this absurdity for once and for all. Please.
- In reference to point 4, the flaw here should have been obvious to all concerned. Defending teams score into an empty net 3 times as often as the trailing team scores the tying goal. That is in a period of 90 seconds or less. If you extend that period to 30 minutes, as this study suggests, the defending team is not thereby still 3 times as likely to score one goal as the other team is to score one goal: it is as likely to score about 20-30 goals. So this idea only works if the trailing team not only scores first, but continues to score, several times, before the other team can score. And most of those goals will be scored before the trailing team, of course, can tie it.
Okay, it’s not all that complicated. If pulling the goalie gives you an advantage, why wouldn’t both teams do it? Yes, I mean at the same time.
And to the rabble who go, “of course the team with the lead wouldn’t pull the goalie– why would they? They have the lead, you moron.”
There– feel better? Got that off your chest? Now let’s proceed.
In every sport, in every aspect, something that is a real advantage for one team is a real advantage for the other: a heavier bat, a bigger goalie pad, height, size. (In football, for comparison’s sake, we’re talking about the defensive or offensive alignments being comparable: not one team’s offense vs. another team’s defense.)
So why wouldn’t both teams pull their goalie for the last 3 minutes of a close game? The naive answer is, because only the team that is trailing would benefit. That is sheer absurdity: an advantage is an advantage no matter what the score is. The naive fan, however, decides that the trailing team will never score in the last minutes of a game if they don’t pull their goalie. Meantime, the defending team would obviously benefit by putting the game out of reach, even though, I will concede, that doesn’t increase the value of the win. But increasing your chance of winning does have a real value. Obviously.
That is not a dispensable piece of information: it is indisputably true– there is no factor that makes it more advantageous for one team to pull their goalie than the other. Either you increase your chances of winning or you don’t. Unless– and here’s the fly in the ointment of those who argue for pulling your goalie early– unless you assert that no team can score in the last few minutes of a game if they have not pulled their goalie. That is plainly nonsense, but it hasn’t been proven because not a single coach in the NHL will not pull their goalie in the last minute of a game in which they have a one-goal deficit. Not one. If a coach, using his head, decided to not do it, he would face hysteria on the level of the Salem witch trials.
And the same with a coach with a one-goal lead who pulled his goalie with a face-off in his opponents end with three minutes left. If this is such a freakin’ advantage, why wouldn’t he? Because, you admit, it would be stupid.
And if it is an advantage because teams only do it if they have a face-off, or possession, in the other team’s end, then obviously the team with the lead will occasionally also have a faceoff or possession in the other team’s end.
Yes, it would be stupid. Because he will have increased the chances of his opponent tying the game with an easy shot into the empty net.
So why is this not stupid for the team that is trailing by one goal?
So team “A” has a one goal lead and a faceoff in their defensive zone. Team “B” pulls their goalie. Suppose Team “A” gets the puck– a not unlikely development — and gets it out to centre ice. (I leave aside the point to be made that a team may begin to adopt the strategy of actively shooting for the empty net; in fact, many teams have clearly already begun to adopt this strategy, which I believe, will kill this entire movement relatively quickly.)
Back to my hypothetical: Team “A” shoots it deep into Team “B”‘s zone and pulls their goalie. So now we have both teams with six skaters and empty nets.
What do you think would happen? Do you begin to understand why this is an absurd strategy? Or, if Team “B” scores to tie the game, do you really think they will leave their goalie out so they can try to win it in regulation time?
Here’s a scenario that, I will concede, makes some sense. A little. A preposterous amount, but some: pull your goalie whenever you have a face-off in your opponents end of the rink.
Well, nobody does that either. For reasons I would think were obvious.
The fundamental problem with this entire discussion is that nobody– not a single coach– will test the theory that not pulling your goalie works better. Not one. Not a single fricken coach. Because the team’s fans would raise hell and every commentator and sports-writer would call him an idiot.
I have noticed that many baseball commentators continue to act baffled when teams don’t automatically sacrifice bunt when trailing late in the game with a runner on first and nobody out, 20 years after this strategy was clearly proven to be worthless.
Note on another factor not measured: it is obvious that a team that is trailing plays differently in the last few minutes of a game– they see time running out. They expend more energy and effort because they know there is an approaching deadline after which they can rest. The know that if they expended that much effort and energy earlier in the game, they would not have much of it left for the finish. This is a bit ritual, and a bit practical. And this is why I would argue that pulling your goalie may actually be counter productive, but here’s the bottom line: we can never know what the chances of scoring without pulling your goalie are, because nobody does it.
(Actually, somebody does: apparently, pulling the goalie is rarely done in the Russian professional leagues. They don’t believe in it. It would be really interesting to get their data.)
To “prove” that pulling your goalie is a smart strategy you must have a baseline to compare it to, just as Bill James had a baseline of teams not sacrificing to compare the strategy to. (It’s easy in baseball: most teams don’t normally sacrifice early in the game, and it’s easy to measure if there is a improvement in the success of the strategy later in the game.) That’s why his conclusions are reliable, and yours, empty-net enthusiasts, are not.
But even without this kind of baseline, listen to yourselves:
Since the best teams in hockey win about twice as often at the worst teams, they are likely behind near the end of the game roughly half as often, and therefore would get only about half the benefit from optimal pulling. If the average benefit is 0.02, we might guess that the benefit is 0.0133 for the best teams….
(From SSRN_Pulling_the_goalie_hockey_and_investment_implications.pdf)
In other words, even if it did have a demonstrated benefit, it would be tiny. None of these studies seem to factor in the possibility of increased intensity of effort in the last few minutes resulting in a normalized base scoring potential that would be higher– which seems likely– than the baseline used to calculate the pulling the goalie conveys an advantage. And, more importantly, none of these studies factor in the possibility that defending teams will behave differently if teams pull their goalie earlier.
We are, in fact, already seeing teams take pot-shots at the empty net. They seem to be more than happy to accept a few icing calls as the expense of this strategy. Think about it: if this happens in the first few seconds after a face-off– precisely when it is most likely that you might have possession of the puck– the risk is substantially minimized.
After suggesting that a team should pull their goalie in the first period if down by 4 or 5 goals or more — seriously?
Though we’d admit the assumptions and simplifications of our model are probably being pushed much harder for this “losing by a ton early” analysis.
Ah– the writer begins to realize just how absurd his premise is.