Are We Wasting the Backup Goalie?

For a long time, teams only really used their starting goalies. Backup goalies existed in case the starter was injured, and the starter got to rest maybe once every eight or nine games. We know better than that now- goalies playing in back to back games in particular are susceptible to injuries and poor performance.

So the modern backup goalie plays to rest the starter on back-to-back games, drastically reducing their chance for injury and helping them come back rested for their next game. But there's a problem. When the starter is "resting," they still need to dress and sit on the bench. If the backup gets hurt or has a terrible game, they're going in, so they need to stretch, prepare and focus as though they're playing, and they might not even get to rest.

Teams are using their backup goalie to rest their starter in the hopes of keeping them healthy and improving their performance. But the current method of rotation is inefficient at this- how could we change it?

A Three-Goalie Rotation

What if, when we rested our starting goalie, we scratched them instead of having them back up? They don't need to eat their special pre-game meals or show up to the rink four hours early to warm up. Instead, they could lounge around all day and show up to the game in street clothes (or even stay home and watch on TV). It would be a proper rest day.

In order to achieve that, we would need to use an extra roster spot. Goalie #1 is the starter, expected to start most regular season games and every playoff game. Goalie #2 is the backup, who plays in back-to-backs or becomes the starter when #1 is injured. Goalie #3 is the all-time backup- they sit on the bench every game, coming into the game if the starter gets hurt or plays terribly.

In addition to letting the starter (and the backup) rest more, there are additional benefits:

• On days where the starter is not playing, they can get sent ahead of the team on road trips, reducing the impact of travel for them
• If there is an injury, it doesn't impact the minutes for goalies in the minors- the prospects can keep playing in the AHL or ECHL, and don't need to get called up to back up
• The #3 goalie can stay after practice on to help skaters get more reps without adding to the workload of the starter or backup
• Switching goalies in-game won't wear down your starter

Seems pretty good to me.

Not every team will be able to do this- it uses up an additional roster spot on a player who is guaranteed to not play much, and that player will eat into salary cap space. There is an argument to be made that having a 14th forward, giving you a scratched center and a scratched winger, or an 8th defenseman, giving you additional depth in a more impactful position, is more valuable than having a fully-rested goalie all the time.

With that caveat out of the way, I want to explore one of those benefits we get from a three-goalie rotation: switching goalies in-game won't wear down your starter. On its face, that doesn't sound too special. If your backup plays badly, you don't need to question whether or not you should pull. That might turn a blowout into a winnable game once or twice a season- a marginal but important improvement.

But I'm thinking bigger than that. Being able to pull the goalie without worrying about fatigue opens up a whole avenue of possibilities beyond just switching goalies because one is playing badly. The bench goalie can become a specialist.

A Closer

We can give our starter even more rest. Let's do some napkin math. Last season, van Fedotov of the Philadelphia Flyers had the worst save percentage of any goalie to play multiple games, with 81.1% (I know raw save percentage is not the ideal way to measure goalie performance, but it's good enough for what we're doing today). The team which gave up the most shots per game, the San Jose Sharks, allowed 2,248 shots against at 5 on 5, an average of 27.4 per game, or 9.1 per period.

A team with the lead gives up more shots, to the point that possession metrics need to be adjusted, so let's say our team will give up 11 shots in the 3rd (8 in the first and second instead of 9 every period). That means, with the worst goalie and the worst defense, a team could expect to give up 2.079 goals in the 3rd period. That awful team leading by 3 or more at the end of the 2nd could take their starter off to give them the night off, and as long as they don't take any penalties, they can expect to win. That would save the starter 1/3 of a game's effort, letting them comfortably (theoretically) play another 1/3 of a game later in the season.

But we don't need to worry about a team that bad, so let's be more optimistic. The worst save percentage from a goalie who played 10 or more games came from Antti Raanta, who posted an 87.2%. The median team last year, the Anaheim Ducks, gave up 1,920 shots against at 5v5. That's 23.4 per game and 7.8 per period. Assuming (and we're doing that a lot here, but this is just a thought exercise) a similar adjustment for the 3rd period gives us 9.4 shots against in the 3rd period on average. With the imaginary Raanta in net, that gives us 1.2 goals against per game. A team up by 2 or more goals could comfortably expect to win.

Let's apply this strategy to a team's season. A marginal team that could have really used some marginal gains. The Detroit Red Wings just narrowly missed the playoffs in 2024, missing the second wildcard spot in the Eastern Conference because of a tiebreaker with the Washington Capitals. If they could pull out even a single standings point, they would have beaten the Capitals and made the playoffs.

They won by 2 or more goals 23 times. Of those, primary starter Alex Lyon played in 10, and of those, 8 saw the Wings leading by 2 at the end of the 2nd, making them eligible for putting in a closer. Putting in the imaginary #3 goalie could have saved Lyon the fatigue of 160 minutes of play (not including any impact from a scratched day off as opposed to backing up), theoretically saving him the wear and tear of two entire hockey games.

If the Wings used Lyon, with a 90.4% save percentage on the season, instead of Ville Husso, with an 89.2%, in just two games, that could have made the difference the team needed to make the playoffs without adding any workload to the starter. Detroit gave up an average of 32.4 shots on goal against per game, meaning that Lyon would be expected to give up 6.22 goals against in those two games compared to Husso's 6.99. Raw save percentage and shots on goal against are not the best metrics for evaluating goalies, so let's be conservative and say Lyon could be expected to save 0.5 of a goal in that time, or a 50% chance of saving an additional goal.

Lyon was injured in December, but there were exactly three games where Husso played, Lyon was healthy, and the Red Wings lost by 1 goal. If Lyon could have played two of those games, there's a 50% (assumed) chance he can pull an OT loss into a win or a regulation loss into an OT loss by saving that one deciding goal. That brings the team into the playoffs.

It's harder to quantify the impact the rest can have without data, but I expect we would see improvements to the starting goalie's save percentage more generally 

A Puckhandler

Maybe the backup can be more useful than just someone you plug in to rest your starter. We want every player on the 20-man roster to be an impact player, someone who can step onto the ice and make a difference.

The goalie position is inherently limited. The mission of the goalie is already so singular- there isn't room for a checking goalie and a scoring goalie, because the goalie is inherently focused on defense. Right?

Here is my idea: when your team has the powerplay, or they're losing and need to score, put in your backup goalie. Recruit the most cartoonish goalie possible, one who can't stop a puck to save their life but who can pass like an actual hockey player. Maybe even give them specialized gear, like different gloves to make that easier. They would come onto the ice, sweep up any dumped pucks, and act like a 3rd defenseman on the breakout. 5 on 4 could become 6 on 4 for the breakout, speeding up transition and leading to more time in the offensive zone.

Since penalty kills get fewer shots, there isn't much risk to putting a complete liability in goal during that time. They would rarely play for more than 2 or 3 minutes at a time, so there's not much risk of the starter getting "cold" on the bench either- players routinely take extended breaks in practice. They would just come on, fling some passes, and come off at a whistle (ideally about 10 seconds before the end of the powerplay). A quality shot-stopper who is also a puckhandler would be expensive, far too valuable to relegate to a #3 position as a powerplay specialist- so we'd be looking for a lower-tier shot stopper to keep costs down for such a low-impact role.

I have no way to quantify this, because it generally matters so little. The very best puckhandling goalies, even when they're regular starters, rarely get more than 3 assists per year. Their contribution to the offense would be only transitory. The question is, would the benefits (however we choose to quantify them) of a puck-playing backup goalie outweigh the negative affects of having a worse goalie to throw in when the starting goalie gets hurt or has a stinker?

Probably. If the team is pulling the goalie because of poor performance, it's generally best to write that off as a loss anyway- having a bad goalie come in to replace won't make a difference. In-game injuries are rare- maladies tend to reveal themselves after the game when adrenaline has worn off. But they do happen. Let's assume our goalie is a similar quality to an emergency backup- there's a small sample, but teams are not destined to lose when it happens.

Let's be pessimistic about it. Last season, the median team allowed 7 shorthanded goals. Let's say our puckhandling specialist allows 50% more goals against, so 3.5 more. Then let's say our starting goalie gets injured mid-game once a season and our specialist allows 2 extra goals against in that game. That means in order to be a net positive, our player needs to add more than 5.5 goals of value. A median team in 2023-24 would have 248 powerplay opportunities (Tampa) and convert at a 21.56% rate (Nashville), which would give them 53 goals. To be a net positive, our goalie needs to help get the team to 59 goals or a 23.79% conversion rate.

So our goalie would need to be good enough at moving the puck to help the team score on almost 3 more powerplays per 100.

Nashville, despite having a completely average powerplay, was very good at breaking out and gaining the zone, 5th in the league last season. Even if every team above them in setup percentage is better at converting their powerplay chances, 11 teams that were worse than them at entering the zone were also better at turning penalties into goals. If zone entry rate directly correlated to scoring, I would say unequivocally "yeah, teams should do that." Maybe there is another, better measure that we should be looking at, like how long the powerplay actually spends in the zone.

NHL Edge tracks that data for us, so let's dig into it.

 

Offensive zone time correlates pretty well with powerplay success rate, but defensive zone time correlates more strongly. Lucky for us, that's where our goalie could theoretically make an impact- by moving the puck more quickly out of the defensive zone.

It's not the strongest correlation, but let's churn the numbers anyway. Our totally average Nashville Predators powerplay spends 25.2% of their powerplay time in the defensive zone. In order to (potentially) get the powerplay conversion rate up to what we'd need, the team would need to spend 24% or less of their time in the defensive zone.

Could a good puckhandling goalie impact the location of the puck 1% of the time? I think so. Should teams be using their backup goalie as a powerplay specialist? I don't know.

An Enforcer

No.

Closing Thoughts

Teams are not using their backup goalie position to its full potential. The player isn't having an impact from the bench, and they're not helping the starter rest as much as they could. I don't know what would be best, and the answer might be different for different teams. I would love to see teams experiment with this to see what's best and squeeze more wins out of every player on their team.