Every year there are a handful of teams that start out at a certain level – good, middling, bad – and are then either impelled upward by glorious goaltending or dragged downward by shoddy goaltending. A very good team can come *this* close to missing the playoffs if it sticks with Ty Conklin; a bad team can ride Semyon Varlamov all the way to the postseason.

Guess which side of that chasm the Oilers are on this year.

We all know how this happened. The Oilers’ roster had a lot of problems and so the team opted to go with a pair of unproven starters with good short-term track records. Sometimes it works and works brilliantly, as it has for Chicago with first Antti Niemi and then Corey Crawford, or as it has for Tampa Bay with Ben Bishop. But sometimes it blows up in the face of the team trying it.

Edmonton went with two guys without long NHL track records. Here are the career numbers for both at even-strength, along with their performance this season:

Oh, whoops. I must have mixed the data up. Those numbers actually belong to Jussi Markkanen and Ty Conklin and point to the 2005-06 season. Let’s try that again, this time including Viktor Fasth and Ben Scrivens. Goalies here are ranked by their career even-strength save percentages:

Two goalies, each with a solid but short-term history, are signed to provide a team with a quality tandem. Both subsequently implode. Edmonton has read this book before, though I guarantee this time the ending doesn’t involve Dwayne Roloson.

There are (at least) two big points with goalies that are really hard for people to grasp – and when I say people here I’m including myself, believe me. One is that performance tends to vary significantly year over year, and it isn’t always team related. Another is that it takes a very long time to identify an individual goalie’s true talent level.

Let’s start with variance. Goalie performance tends to fluctuate even when the team’s defence stays more or less unchanged. It’s how Corey Crawford can post a 0.917 SV% in 57 games for Chicago one year and then post a 0.903 SV% in 57 games for Chicago the next year (goalie apologists don’t typically go after Chicago’s defence because they know how implausible it sounds). It’s how Mike Smith can post a 0.930 SV% for Arizona and then drop 20 points the very next year. For that matter, it’s part of the reason Devan Dubnyk can go from being a guy with a save percentage between 0.914-0.920 for three years behind the rebuild Oilers and then suddenly collapse in Year 4.

I’m using “variance” here to capture pretty much everything we can’t account for with our current data, which is a lot. From nagging injuries to personal circumstances to goalie coaching to mental state to plain old luck (and there’s a lot more random chance than people really like to admit in hockey) there are all kinds of factors that go into a goalie’s performance beyond a) his pure ability and b) the defensive performance of the team in front of him.

Humans like simple stories that explain all those things, and a lot of times we like those simple stories so much that sometimes we make them up. The fact is that a lot of times things are going on which we don’t know about, and for outsiders the best way to handle it that I’ve seen is simply to note that performance is unpredictable game-to-game, month-to-month and year-to-year.

The shorthand: Henrik Lundqvist can go from being a 0.888 SV% goalie in December to being a 0.938 SV% goalie in January, as he did last year. I don’t know why that happens, though I’m content to say his defence probably didn’t get 50 points better in the span of a month. Massive fluctuation is just the nature of the position, and people need to accept that.

So why don’t we worry about those fluctuations with Lundqvist? They can be pretty bad, after all – that 0.888 SV% run last year came on 260 shots, which is three-quarters as many as Fasth has seen this year. Going back to 2011-12, from March through April Lundqvist was a 0.895 SV% goalie on 399 shots; that’s a pretty substantial chunk of the season.

We don’t worry about them because we know what Lundqvist is. He’s been a 0.920 SV% goalie on nearly 17,000 shots over his career. There have been some peaks and valleys along the way; they balance out and we know he’s quality.

It takes a lot of shots to be sure of a goaltender, though. Brian Macdonald (now employed by the Florida Panthers) did some great work on this front, and passed along his estimate of underlying talent to SBNation’s Eric Tulsky (now employed by an unknown NHL club) in a piece that ran at SBNation.

Imagine a goalie to had posted a 0.925 SV% (better than Scrivens entering this season) on 2,500 shots (more than Scrivens had faced this season) at even-strength over his career. That’s five points better than the 0.920 EVSV% league average of the last few years; even so Macdonald’s work suggests that this goalie’s true talent could fall anywhere from 0.916 to 0.932 – a gap greater than the observed difference over the last seven seasons between Lundqvist and Dan Ellis.

Goalies are always gambles; there’s just no getting around that. It’s extremely hard to build up the kind of sample necessary to nail down true talent and the peaks and valleys along the way are maddening. Sometimes things start badly and snowball from there and the only solution is to go out and get a new goalie because the guy between the pipes needs a long break to get his game back together.

Edmonton’s there. Watch the video:

Shot from an impossible angle somehow finds the five-hole; no pass involved.

Straight shot from outside the scoring chances area; no pass involved.

Goalie bites so hard on the initial shot, which is blocked, that he’s in no position to stop a shot from an impossible angle; no pass involved.

Short side from the very far corner of the scoring chances area.

Those four goals, all ranging from iffy-to-terrible, are all from the last three games, two of which have been one-goal losses. The Oilers have plenty of problems, to be sure, but this is a team that of late has been average one goal per game on a non-scoring chance shot (the fourth was certainly stoppable but not nearly as odious as the first three). NHL teams lose, and lose often with that kind of goaltending.

Scrivens and Fasth can certainly be better than this; they’ve shown that repeatedly over their (short) careers. The question is whether they can rebound in Edmonton before the end of this season.

No single issue is killing Edmonton the way goaltending is. Since trading for a new guy doesn’t seem to be an option the team is interested in, the Oilers will just have to hope that one of their goalies can find his game on the other side of the Christmas break.