To pull or not to pull Andrei Vasilevskiy

The goalie is pulled, an empty net goal is scored = bad decision.

The goal is pulled, a tying goal is scored = good decision.

Of course, this is simplistic thinking and we’re better than that.

We’re smart enough to know that it’s not wise to confuse an outcome with a decision. Certainly not when it comes to the multi-variable universe that is a professional hockey game. There are just too many unknowns, too many uncontrollables; there is no clear choice that will lead to success. I mean, if winning were easy every team would always do it.

To be down a goal with minutes or seconds left in the third period is to be in a tight tough spot no matter what you do. The longer the game remains that way, the less likely you are to win that game (or even send that game into overtime when you can get a point anyway in the cockamamie system the NHL uses to decide games tied at the end of regulation, a system in which one team wins and the other doesn’t lose). Every single tick of the clock without a goal in your favor reduces your chances.

The Tampa Bay Lightning are in an interesting situation with Andrei Vasilevskiy

By now, Tampa Bay Lightning coach Jon Cooper has a reputation in late-game situations of this kind. He pulls his goalie first and asks questions later. In other words, “Cooper is always aggressive.” This is the sort of utterance you are to hear if you watch or listen to a broadcast. It’s also the mantra if you question his decision after the game in, say, an X feed.

Comments from the largely team-sponsored and team-adjacent media — as well as many partisans partial to a coach who has had more than a little success in a stellar 12-year tenure behind the Bolts bench — go like this: “It’s what he always does.”

This is meant as a defense. What it is, actually, is an indictment.

Always? As a coach, do you want to do anything always?

A hockey game isn’t a game of Black Jack where you play certain odds and if you remain disciplined you come out ahead a little more often than you lose. Hockey is about humans playing cards and sometimes the hand dealt is seeming in your favor and other times not. You’ve got to read the room. The only relevant data can’t be that which was found in a spreadsheet before the game.

How is the team playing in this game? Against this team? Lately? That’s just for starters.

It used to be that teams down a goal would pull their goalie at the last minute of play or thereabouts. In recent years, thanks to analytics, teams have largely followed the numbers that say that’s too late. So it’s not unusual for a coach to pull the goalie sooner, often much sooner. Cooper regularly pulls his goalie with more than three minutes to go in the game.

He did so in each of the two games heading into Tuesday night’s tilt in St. Louis. In both of these games, the other team scored an empty-net goal and the Lightning lost. In St. Louis, Cooper pulled his goalie with fewer than two minutes to go, down 3-2.

Let us consider these examples as we talk about the larger decision. For, alas, it’s unlikely to be the last time the coach makes a similar one this season.

Let’s concede that when down a goal near the end of a game, it’s a decision of when, not if, to pull the goalie. To have the clock run out at the end of a one-goal game and your goalie is standing in his net would be to tell your team you’re not trying hard to win.

So when is a good time? Among the factors a coach has to consider: personnel and puck control. Ideally, you want your best scorers — probably your top power play — fresh and on the ice. You also want the puck to be under your control or, short of that, soon to be dropped in an offensive-zone face-off. Since this is the only time in which you play six skaters, there is also consideration as to who the extra skater will be.

In other words, the decision to pull a goalie can’t rest solely on the game clock. When your players are rested and in control of the puck are not things a coach can control; Cooper can’t wave a wand. (If he could, presumably he would have done so before his team was down a goal late in the game.)

If a coach waits too long, his team might not have much chance to play with the man advantage that comes with a sixth skater. A minute in a hockey game can go by in a blink.

Yet, there seems to be a threshold here. Circa 2024, an NHL player with the puck on his stick in space outside his own zone is going to deposit the biscuit in the basket almost every time. How long are your best scorers going to be able to control the puck? How long are they going to be able to stay on the ice? Three minutes and the odds are really high you are going to give up an empty-netter.

On Friday at Minnesota, down 3-2, Cooper pulled Andrei Vasilevskiy with more than 3:30 left. I happened to be in the arena. Without thinking, I said it out loud: “This is too soon.” Almost immediately, the Wild got control of the puck and it was 4-2.

I’m no savant. There’s a reason I am not paid to make such decisions for a professional hockey team. What I am pretty good at is feeling. This just didn’t have the feel of a moment in which desperate measures — make no mistake, to remove your goalie is to throw a Hail Mary — were called for. I will add that the Lightning players on the ice looked tired; so that being the ideal time doesn't add up.)

The Lightning had outplayed a very good Minnesota team on the road in the first period. The Wild were the better team in the second period. But it was still 1-1 entering the third and Tampa Bay had reasserted itself. The Bolts were easily out-shooting the home team and, offensively, they have a number of players off to potent offensive starts this season.

Nick Paul is scoring. Brandon Hagel is fantastic. Brayden Point was on pace for another 50-goal season, at least before he was injured on Sunday. Nikita Kucherov wasn’t having his best game of the season, but with Connor McDavid sidelined with an injury, he was, at that moment, the most lethal offensive player on the planet. I would have liked to see the game play out a little longer.

Sure enough, after the Wild made it 4-2 the Lightning did score. This might lead credence to my view but even I won’t claim that it does. The fantasy-ization of sports is such that we think that the players are like pieces on a chessboard. Remove one player or play and the rest of the board and the possible moves that could be made remain the same. No way.

A coach makes a decision like whether or not pull the goalie and a goal is scored and nothing after that can be assumed to have occurred before that decision. Absolutely nothing, not even the way the puck slides in a particular spot on the ice, which has been cut — or not — by the moments that just occurred (or didn’t). Maybe the Wild relaxed after taking a two-goal lead. Maybe the Lightning would have tied the game without a sixth skater. We. Don’t. Know.

But, again, the decision looked, well, not right. There was every reason to think the Lightning might make a push. Maybe in the frenzy they draw a penalty. We’ll never know; we don’t get to prove counterfactuals.

Sunday afternoon in Winnipeg a similar scenario played out. This time, the Bolts were down 5-4 and Cooper pulled his goalie with a little over three minutes to go. It still seemed early to me, but I was slightly less offended. The Lightning were playing backup Jonas Johansson who is only slightly better than an empty net. (I’m kidding! But the Swede had given up a handful already.)

On Tuesday night in St. Louis, as I said, Cooper kept Vasilevskiy in the net until about 1:51 to go in a 3-2 game. Which is downright conservative for him. Why did he wait so long? We won’t know because no one in the media will ask him. He had a chance 30-plus seconds before when the Lightning had control of the puck with zero pressure. Maybe the coach simply couldn’t avoid recency bias. Incidentally, neither team scored and so there’s that.

All told, three moments with two aggressive decisions and one moderate one. All losses. To me, the St. Louis approach makes the most sense most of the time. We shall see if that call augers a change in approach from the coach (I wouldn’t bet on it).

As the clock counts down in the third period simultaneously, so do the odds of winning a game in which you are trailing. When to pull a goalie is a judgment call; it’s not an exact science. No doubt, analytics cuts through human biases. Data is important. It’s just also the case that teams play out the last minutes of a given game one at a time. Odds compiled over seasons by even the same organization are only partially relevant to a given moment at hand.

Of course, it’s less than ideal to find yourself down a goal late in the game. The Lightning has been making a habit of that of late — three straight games heading into Thursday night's game at home against the Philadelphia Flyers — and, regardless of any single coach's decision, that’s a bad data point.