I don’t think there is a single roster spot in professional hockey that is more over-discussed, and more over-loaded in arguments than the backup goaltender position for the Toronto Maple Leafs.

Naturally, I’m going to contribute to this by discussing it further and giving you even more fuel to argue with. But the objective here is not to come out with a firm conclusion, but rather reset the parameters. Ultimately, I think it’s lost on people that we’re dealing with a lower-leverage, lower-priority roster spot here, and a lot of that is because of the unique circumstances that have come from the past several years in the role. So let’s dive into it, bit by bit.

How Did We Get Here?

Truthfully, the Leafs have almost never had great backup goaltending in the salary cap era. For the first decade or so, the Leafs could be relied upon to run a goaltending tandem of “incredibly inconsistent and/or unhealthy starter” as the 1A, and “the last guy we tried this with” as the 1B. This fact typically masked the concern about the backup, since getting a starter we’d all be happy with was a much higher priority.

That changed in the summer of 2016, when Lou Lamoriello acquired Frederik Andersen from the Anaheim Ducks. The cost was steep – a first and a second-round pick is an extremely premium price to pay for a goaltender in today’s market, despite the importance of the position, and from a relative standpoint it could still be argued that Toronto overpaid.

But ask any Leafs fan today about the cost and they’d say “yeah, so what?” and you can’t really blame them. Andersen has given the team consistency between the pipes like few others have in modern history, stopping an above-average amount of pucks, rarely going on extended slumps, and playing over three-quarters of the available games. While things may change when his contract ends, Andersen appears poised to give the Leafs at least a half-decade of peace of mind for the majority of their season.

Behind him, though, has remained a rocky battle. Toronto first tried Jhonas Enroth, a goaltender never really given a starter’s look due to his size (standing at just 5’10 in the tallest era for goaltenders ever), but known as one of the most dependable backups of his generation. He lasted six games under a very disappointed Mike Babcock, who made it explicitly clear that he wanted a goalie who filled more of the net. Enroth’s NHL career was effectively ended by that stint, though he’s fared well in the KHL as a starter for Dinamo Minsk since.

The Leafs found themselves in limbo for a few weeks, sliding Antoine Bibeau up a couple of times in relief while trying to figure out who was next in line. Eventually, they settled on claiming Curtis McElhinney off of waivers, known as one of the league’s worst regular goaltenders (with some statistical models having him as the worst). The move was seen as the time as a desperate Hail Mary. But something happened.

Building on a recent streak of decent form with Columbus, McElhinney closed out the year as a shockingly dependable safety valve for the Leafs, culminating in a playoff-clinching save on a Phil Kessel pass to Sidney Crosby in Game 81. He became a cult hero for a streak that saw him beat the odds, and that was barely scratching the surface.

In the next season, 34-year-old Curtis McElhinney, of a career 0.905 save percentage heading into 2017/18, put up one of the best seasons that anyone has ever had under a backup’s workload. His 0.934 save percentage was tied for the sixth-highest ever for a 10-20 game goaltender, and was the also the sixth-best for any goaltender over the age of 30. His comparables became names like Tim Thomas, Dominik Hasek, and Jacques Plante, albeit in a much smaller sample.

That combination of small sample and age meant that the Leafs were grateful for his efforts, but unwilling to throw away their succession plan the following year. Rather than stick with what worked, the team opted to promote 25-year-old Garret Sparks to the big club. Sparks had come in as injury relief and tank fodder in the 2015/16 season, but this was his first scheduled appearance in the roster – and one that followed a 2017/18 season with the AHL’s Toronto Marlies that was its own kind of historically dominant – a 0.936 SV% in a starter’s workload during the regular season, and a run to Toronto’s first minor league championship in the playoffs. McElhinney was placed on waivers in hopes of clearing him down to the Marlies, but was claimed by Carolina.

This caused a relatively big divide in the fanbase – or, rather, became a battleground for one that already existed. McElhinney was a Babcock favourite and his veteran experience endeared him to the more “old-school” side of the fanbase, while younger fans and the more stat-savvy crowd leaned towards betting on the prospect over the hot-hand, good-soldier veteran. Every night each would post on their respective team was heavily scrutinized, and for a while, things looked pretty neck and neck statistically, with both putting up fantastic results heading into this time last year.

Unfortunately, a combination of scrutiny, misfortune, and a pair of concussions suffered from errant shots in practice slowed down Sparks, and having to step into a 1B workload to fill in for Petr Mrazek during an up-and-down season in Carolina slowed down McElhinney, though the latter ended up with the better overall season.

A combination of Babcock’s lack of confidence in another backup goaltender, the low approval rating within the fanbase, and the fallout from a not-untrue but fairly needless comment from Sparks in March led to his departure in the summer, meaning Michael Hutchinson would be next to take the reins. Hutchinson had produced a better season save percentage with the Leafs than both Sparks in Toronto and McElhinney in Carolina in 2018/19 (0.914), but didn’t have much of an NHL body of work, and had been terrible to start the year with Florida, so there were plenty of questions coming into the fall from people unsure that lightning could strike twice.

Which leads us to now. Hutchinson is near the bottom of the league in save percentage, (0.876 in 7 games), and has gone 0-5-1, turning the once-and-for-all “drama-free” pick into a bit of a white flag, with many wondering where exactly you go from here.

Balancing The Budget

A lot of the backup goaltender debate centres around an opportunity for people who question Kyle Dubas and the current Leafs management group to point fingers at them, because it exploits two potential avenues for criticism.

The first is process over results, which is what one could simplify the Sparks over McElhinney decision to. Both netminders carried risk of failure, but one carried more potential gain: picking Sparks and having everything go right would have meant a goaltender that you could ease from a backup to a 1B to a starting successor, whereas a hit on McElhinney would have meant slightly improved results for a couple dozen more games at most, even if he never regressed and played the rest of his career in blue and white (McElhinney is currently in his Age 36 season). Many, however, felt that time was of the essence, and also didn’t recognize that continuing down the road they were already on was still a risky bet – this wasn’t a scenario of a safe play versus a risky one.

The second, which we’ll cover here, is the budget. The Leafs’ handling of the second contacts for their big three kids (Matthews, Marner, and Nylander) could have been better, particularly in terms of setting public narratives and has pushed the team towards a “stars and scrubs” model of roster composition.

Personally, I’m a believer in that model, though not everyone is. Whatever the case, it led to the Leafs’ budget for a backup goaltender being somewhere near the league minimum, which has led to the assertion that the Leafs painted themselves into a corner with Hutchinson.

First and foremost, the easy response here is that I would rather have a core skater than a backup goaltender, who won’t even play in the vast majority of my games and ideally, will never see a second in the postseason.

Next, you have the marketplace. The list of backup goaltenders who made in excess of $999,999 by switching teams as a UFA this summer is as follows:

  • Keith Kinkaid (1 x $1,750,000)
  • Curtis McElhinney (2 x $1,300,000)
  • Mike Smith (1 x $2,000,000 w/bonus cap of $3,750,000)
  • Semyon Varlamov (4 x $5,000,000)
  • Sergei Bobrovsky (7 x $10,000,000)

Toronto was obviously not in the market for a second starter, knowing that Andersen gives them some of the best dependability in hockey for 60 games a year, so that rules Bobrovsky (who has been DOA in Florida) and Varlamov (who has been fine, but may have already lost the starter role in NYI) out. Smith is having a better year than he had last season, but most would be disappointed at a 0.907 SV% for a $2,000,000 backup, given how people reacted to Sparks putting up a 0.902 for $675,000. Kinkaid has been worse than Hutchinson this year at nearly triple the cost, and cleared waivers this afternoon.

This leaves a McElhinney return as the only missed door here, which is kind of fitting.

The reality is, though, while McElhinney showed he was good enough to have held the fort last year should the Leafs have kept him, he also saw his save percentage drop 22 points from 2018 to 2019. Looking at his history, it’s pretty evident that 2018/19, along with a few games prior and following, was the absolute peak of his career. Giving a 35+ contract above the buriable threshold to a player falling backwards, with father time not on his side, would have still been a bad idea. While still outperforming Hutchinson, McElhinney has barely been above the league average this year and is continuing to slide back down, meaning that while he would have exceeded the F- performance Toronto is currently withstanding, it would not have been a universally agreed-upon decision, nor would people be universally happy with the results.

Odds are, being limited to that $700,000 or so budget did not prevent the Leafs from adding a safe bet backup goaltender, as the more expensive options were either looking for bigger workloads or haven’t particularly dominated. To illustrate this point further, let’s work backwards.

Let’s set the bar for a dependable, satisfying backup save percentage at a flat 0.910 SV%. That’s actually about the league average including starters over the past two years, but many still remember the early-decade years of 0.915 being the bar for goaltenders in general. So I went and found all the goaltenders who have put up a 0.910 SV% at least three times in the past five years.

In total, 42 goaltenders fit this bill, 39 of whom are continuing to play in the NHL this year. Roberto Luongo retired due to wear and tear from years of Hall of Fame play, Scott Darling is continuing a shocking downward spiral in performance over in Austria, and the third disappearance, Michal Neuvirth, was actually offered an opportunity with the Leafs to battle for a backup spot at training camp this summer, but showed up out of shape and uncommitted to turning that around.

But let’s look at what the ones who are playing are up to.

What you’ll notice is that the average goaltender who can consistently put up “dependable” numbers is playing a starters, or at least a 1B’s workload, and is being paid accordingly for the service. Only five of the 39 goalies have played fewer than 10 games so far this year (McElhinney, Kinkaid, Ryan Miller, Cam Talbot, and Jake Allen), and just as many make under $2,000,000 (McElhinney, Miller, Kinkaid, Darcy Kuemper, and Juuse Saros).

This is because, well, they’re better goaltenders, and consistency gets you starting jobs. Too many people look at the starter-backup relationship like baseball, with one game in the Majors looking like a week in the NHL. They see starting goalies as starting pitchers, and backup goalies as relievers or closers. The difference is that pitchers are specifically developed towards those workloads, whereas goaltenders are all developed towards the end goal of being a starter. Some goaltenders will move themselves to a mental routine that is in line with a backup schedule as their careers progress, but that’s typically after proving that they aren’t good enough to play more games.

In other words: The idea of a cheap, consistent backup is by and large a myth. If you’re consistently putting up results, you’re likely more than a backup, unless you come with a very unique caveat. You’re also often making a salary that reflects that, and all that consistency doesn’t even guarantee future consistency – as seen by the fact that half of our group here is currently below the bar they’re noteworthy for regularly clearing.

Origins Of Backups

Another way to evaluate the goaltending landscape, rather than splicing for results, is to do so for usage (sorry for all the binning, Model Twitter). Let’s take a look at the last five years of goaltenders who have played somewhere in between 10 and 20 games played. Note that, for this, I’ve taken out a couple of cases of obvious starters stuck in that usage threshold due to long term injury in a given year (Carey Price, Jonathan Quick).

So typically, these backups to true starts are playing 16 or so games over a full season, putting up a shade over a 0.900 save percentage, and make under a million bucks – often under $800,000.

Now, you may be saying – 0.500ish record, 0.902 save percentage, GAA of about 3; that seems pretty familiar, right? Well, it is – it’s basically what Sparks (8-9-1, 3.15, 0.902) put up for the Leafs last year in 20 games played. He also did it in his Age 25 season, which is right around the median and average age of about 26. In fact, of the 64 seasons that fit this description in that half-decade, he ranks 32nd in save percentage, and 30th in age, at the 16th-lowest cap hit.

What the Leafs got out of Sparks last year was basically the definition of average backup goaltending behind a capable starter. The reason that it didn’t feel that way was because the season prior included an extreme outlier – the fourth-oldest player in this pool of 64 putting up the 3rd-best results, only behind Darling (who Carolina tried to make their starter before his decline), and Antti Raanta (who has stopped tons of pucks for Arizona since).

Today’s pure backup goaltenders tend to have two origin stories. They’re internally-developed goaltenders who are either on the way to becoming a starter or are cost-controlled enough to keep around, or they’re hail-mary, low or no cost acquisitions without a consistent track record, where you hope that your team is where things will be different and/or more steady. Again, you are not looking for the equivalent of a five-star closing pitcher, but rather, a starter that you don’t trust to play more than a quarter of a season.

This means that in these past few years, Toronto has had a failed dice roll, followed by a successful dice roll, followed by a dead average result out of someone brought in the traditional way, followed by what’s looking like another failed dice roll (two if you count Enroth).

To be honest, that’s not terribly bad, or terribly unusual. Teams with top starters go through this all the time, because it’s a low-importance, low-upside position when you already have the bulk of the year covered, and your money can best be spent elsewhere. It feels worse with Toronto because the successful dice roll was a generational anomaly, which made the average result that followed feel like a let down, and the current netminder has started the season with shockingly bad results. It’s more about the order in which it happened than it is what happened.

Timing Is Everything (Or Nothing)

Another element that has come into question over these few years is how it is decided when the starting and backup goaltenders are played.

Under Babcock, a healthy duo typically meant that Andersen would play all non-consecutive games, with the backup playing the second game back-to-backs and games where an injury or illness required them to.

This isn’t particularly abnormal behaviour, as it’s the best way to get as many games out of your starter as possible while still maintaining their health. It means they don’t have to play the full schedule and removes the games where they would feel the most tired.

What people grew to not like was the fact that it was always Andersen in the first half, and the backup (Enroth, Bibeau, Calvin Pickard, Sparks, McElhinney, or Hutchinson), in the second half. I do not use “always” loosely here: In the 36 occasions in the Babcock+Andersen era where he split his goalies up for the back-to-back, Andersen started the first game and the backup the second for every single one.

Perhaps to show a willingness to be flexible, Sheldon Keefe reversed the order for the first back-to-back this previous weekend also pointing out the factors involved (most interestingly: that it was a 4 PM start and that Hutchinson was already used to afternoon starts from recent AHL play). It worked about as well as every other attempted did, with one regulation/overtime win out of the two (for the purposes of what is to follow, let’s call shootouts a coin toss) in 25 of the now-37 attempts at this.

Of those 1-win duos, 18 wins came from Andersen, and just 7 from the backups. Sparks was the best at getting revenge wins, putting up 4 to McElhinney’s 2 and Bibeau’s 1. On the other hand, McElhinney was best at doubling up, with four wins to create 2-0 runs to Sparks 1 and everyone else’s 0 – which probably contributes to why it felt like McElhinney’s appearances were less of a “let down” – maintaining a winning streak can often seem more important from a backup than ending a losing one.

In an attempt to see if going the opposite way would have helped out the team, I flipped the save percentages from each respective night onto each other’s games and checked to see if the new goal differentials would have helped the Leafs win games. So if, say, Andersen posts a 0.950 on Saturday and the backup a 0.875 on Sunday, this flips the two around to see if the scores would have led to different results. It’s crude and ignores some intangible factors, but it at least gives us a general idea if the criticism for not going the other way is fair. To maintain accuracy, I removed all empty-net goals from the totals, and margins between 1 and -1 were pushed to overtime and assigned a point value of 1.5.

Using this method, I found that Toronto would have gained points by swapping save percentages on back-to-backs ten out of 37 times, but lost points this way 14 out of 37 times. Overall, the estimate is a loss of 7.5 standing points if the games were reversed.

This makes intuitive sense to me. If you’re going to make a firm rule for back-to-back management, “starter first, backup second” is a pretty reasonable one. Your team is almost always going to be the underdog on the second night due to exhaustion and a backup not reaching expectations can often mean that you’ll lose by 4 instead of 2 or 3, rather than costing you a win. In the six games that decisively go from loss or shootout to a win if Andersen’s previous night was there in its place, Andersen posts a 0.966 save percentage or higher in five of them, so it speaks more to his performance than the backups.

I do think there is room for flexibility, and I feel like we’ll see more of it now. The biggest factor to me should be standings leverage – there is no reason to go starter first, backup second, for example, if your first game is against an out-of-conference opponent and your second is against a divisional rival. This week’s pair, for example, is a good time to stick with Andersen first, Hutchinson second, because it’s worse to drop points to the Flyers than it is the Avalanche.

I’d say over the course of the year, you’d likely want to do Andersen first this week, first on the 27th/28th (NJD/NYR), second on February 7th/8th (ANA/MTL), 1st on March 5th/6th (LAK/ANA, though they’ll probably reverse it as a former-team gesture to him), and 1st on March 25th/26th (TBL/CAR).

All the same though, these are pretty marginal gains, and it’s nearly impossible to truly point to the results that your sub-20 game backup puts up as season-defining unless they’re historically good, historically bad, or your season ends up relying on a single goal. For example, many pundits have placed blame on Sparks’ performance for costing the Leafs home-ice advantage in the playoffs last year, but given the seven-point difference in the standings between them and Boston, a 0.940 or so save percentage would have been needed to reliably flip enough losses into wins.

As we established earlier, a 0.940 performance from a 25-year-old backup over 20 games would probably make that goaltender not the backup any more and is an unrealistic expectation. That all goes to say: First Half Starter, Second Half Backup has probably helped the team more than it’s hurt, and it’s not the cause for concern, though having some leeway now under Keefe is a good thing as well.

What Now?

Over the course of this piece, we’ve established the Leafs’ recent history with backup goaltenders, why backups are backups, how backups become backups, why the Leafs didn’t budget themselves out of a better option, and why their scheduling wasn’t the problem.

That obviously doesn’t change the fact that they currently have a 0.876 veteran backup who can’t win the team a game, though.

There isn’t really a safe solution here. Currently, there are 20 goaltenders in the NHL who have played 5-10 games this season, which is about where I’d put the backup line at this mark of the year. Unfortunately, the top of the list is filled with 1A/Bs (think Halak, Allen, Raanta, Reimer etc), “goalie of the future” prospects (Jarry, Samsonov, Demko, etc), and guys on benders with contenders (Who is Pavel Francouz?).

One potential option if the price is right might be to buy in on a prospect, especially if the price is right and you’re not sold on Kasimir Kaskisuo, Joseph Woll, or Ian Scott just yet. One who might be intriguing is Alexandr Georgiev with the Rangers; at a little under 24 years old, he doesn’t seem destined for stardom but looks well on track to be a reliable 1B-type, and with Igor Shesterkin looking to be the heir to Henrik Lundqvist’s throne, they may be willing to move their in-between guy.

Another idea is to try again with a bubble-age, bubble results goaltender like they did with Sparks last year. Vegas currently has the market cornered on those, as they decide who they like better between Malcolm Subban and, well Garret Sparks. As much as I think trying one more time with Sparks now that Babcock is gone might be interesting, I could understand the hesitance there. Subban might be the better fit from a PR standpoint.

Next is the blind dice roll, similar to McElhinney’s arrival. Just claim the next vaguely interesting veteran with a low-risk contract (read: not Kinkaid) off waivers and hope it works. This is the least likely to be effective, but also the easiest.

But maybe the best move, for now, is to sit down and take a deep breath. I don’t think Michael Hutchinson is a long-term fix, but he’s very clearly in the midst of the worst slump of his career, and I’d imagine that most of the damage has been done. If he is even just a bad backup instead of a historically bad backup for the rest of the year, the dent that causes should be insignificant in the grand scheme of things as long as as the rest of the team performs consistently.

I’d like him to be better, I’d like to have someone better than him, and the Leafs no doubt feel the same. But after three years of making the Leafs backup job into the highest-pressure goaltending position in hockey, maybe we need to scale back the urgency, realize how small of a cog it is in the machine, and worry about the things that can be controlled until another long-term solution becomes readily available. The sense of urgency has already squashed one decent-enough solution – the next step isn’t to do something desperate for a non-guaranteed chance at marginal improvement.

---
Thanks for reading! Hopefully you enjoyed this post. If you did, don't hesitate to share it on Twitter or Facebook; having more readers will help the site grow. As well, consider a subscription if you're interested in reading additional work that isn't available to guests.