It’s just one game. I feel like I’ve said that before, and there might be a case to be made that you really, really had better not judge any goaltender on one game.
In a minute I’m going to get to looking at goals above expectation, but first we should talk about some other expectations: yours, Mike Babcock’s, the media’s, and everyone else’s.
Variance is what you should expect
You should expect variance in performance from all players, but you absolutely should expect it in goalies. And yet what we all seem to long for is consistency. If everyone understood that variance is normal, the sobriquet “October Freddie” would never have been coined. Variance, much like regression, doesn’t come on a schedule, but people love to see patterns in everything. We like to imagine the chaos is controlled by something, even if it’s only the calendar.
In one experiment a teacher asked his class to make a fake set of results for a coin toss. Heads - Tails - Heads - Tails. In his class, full of people who should know better, so many did such a bad job of faking real random results, an actual trial was done to see if people in general are bad at this. We’re bad at this.
The takeaway is that in any random series of results within the normal expected range has clumps of good and bad, heads or tails, high or low save percentage. October Freddie is just Freddie, just like you’ll flip that coin and see heads five times in a row some of the time.
I like looking at Cole Anderson’s visualizations because they always drive this point home. Let’s look at one from his projections for this season, not least because it’s hopeful:
Bear in mind, this chart is made with Curtis McElhinney as the backup, not Sparks, but the backup has a low impact on things in Leafs-land.
What you see there is the total goals prevented above what an average goalie would save, this is called GSAA. It’s called goals saved over expectation, if you use an Expected Goals model to set the baseline for the average goalie. This is one very simple and readily available way to look at goalie performance that is not all-situations save percentage, which is largely useless for anything but judging how a goalie is perceived.
The full width of the coloured lump (technical term) for each team shows the possible range of projected results. The height of the lump at any one place indicates how often that sort of result is probable. In the middle is where most of the goaltending for that team should lie. But not all of it.
You should readily see that every single team in the NHL will likely experience goaltending that no one would call consistent. Some more than others. Some teams with good backups who are more consistent than their fellows show up on this list with high peaks, sometimes two obvious peaks. Florida, Nashville, San Jose are the three teams that always come to mind for backup quality, with Colorado now joining that group. But you should expect the backup to be less consistent and less good than the starter.
If you read up on the methodology for the projection, there is a section precisely detailing how a 25-year-old rookie gets added in as he produces results:
If a 25-year old rookie is brought up from the AHL, we will probably expect below average results, say an extra goal against every 100 shots (-1%). If they post a 30-save shutout in their first game, the evidence (the shutout) wouldn’t necessarily overwhelm the prior, so combining our prior beliefs and evidence (realized save percentage) into a posterior (posterior probability distribution, blue line below), our updated estimate of their results will better than the prior of -1%, but not by much. However, after 10 games of superb results will begin to move into positive territory.
That’s the method used in this model, but is that the method we use in our minds when we look at a new guy? Is it what Mike Babcock does?
You ask Mike Babcock to wait for 10 games before he decides on his backup
It’s neither you nor I who has to ask that question, rather it’s Kyle Dubas, but as much as fans have cast Dubas in the role of caped crusader righting the wrongs perpetrated by the old hidebound man behind the bench, I’m not convinced it’s reasonable to ask any coach to wait for 10 kicks at the can by a backup before he even starts to form an opinion.
Jhonas Enroth didn’t get that many; he got six. If we put on our hindsight goggles, I’m not sure many think he should have had more.
Mike Babcock is running a hockey team, not a double blind study so anyone, even his own GM, can get a big enough sample to analyse the backup goalie. You don’t get to go back and replay those games if you decide two months later he’s a bust, and I can see why a coach does not want to have his skaters thinking about who is in net.
We don’t really understand the psychological inputs into game results. We think score effects might be influenced by how the coach and the players feel about being ahead or behind in score, not just what their tactics on the ice are, but we’re just guessing.
The guy who has to guess about that is the coach, and if you can come up with a convincing argument for why he should stand behind the bench and watch a goalie be terrible game after game after game, I’m all ears. And no, “he deserves a chance” isn’t convincing. Backups should be seen as expendable, interchangeable, low-importance players unless he’s your future starter.
Ray Ferraro is the guy who likes to say that when you get an opportunity to play, you have to seize it, you have to produce.
Micah McCurdy likes to say something along the lines of: All teams need a plan in place at the start of the season in case their goalie underperforms by 2%. And then they need to act on it.
That’s great advice, but hard to enact for goalies with the waiver rules being what they are. There’s always extra skaters, but carrying three goalies really is a bad idea unless you’re protecting a genuine prospect.
We aren’t at the point where Sparks can be said to have failed at seizing the day, nor are we at the point where the envelope that says Plan B on it needs to be opened. I’m not really sure whose picture is in that envelope, by the way.
If October Freddie is just Freddie, then that Chicago game was just Sparks. And while I find the goal by goal excuse making of his ardent fans a waste of time because it isn’t even all that funny, I am troubled by some of the times he didn’t have a goal against yet looked totally at sea. However, we have no idea of the shape of Sparks’ lump of expected results, not really. So fan or doubter — we’re all just guessing at this point.
Unlike many, however, I don’t wholly discount his past NHL results. Yes the team was bad, no they weren’t that bad, and if what McElhinney did years ago on some other team should make you doubt him now, the weeks Sparks spent driving the tank should make you wonder about him now as well.
I remain displeased that Sparks was never played last season in the NHL, that Calvin Pickard only played once, and that neither played much this preseason. I understand why a coach would do all of those things. They are much more in-the-moment thinkers than management is, but it’s a contributing factor to how we ended up here, and Babcock’s past stubbornness is compounded by his regular-season priorities in ways that will make it hard for Sparks to overcome this early bad performance. The leash is shorter now than it was, in other words.
Maybe what’s inside that Plan B envelope is just one word: Wait.
There are three sets of back-to-back games in November. The first pair is one month from now. That’s a month of practices for Sparks, a month of nothing but Andersen in net, a month of no one having to even consider if the backup situation is a problem or not.
It’s also a month to allow the teams that claimed Calvin Pickard and Curtis McElhinney to find them surplus to requirements again, and for there to be a chance for the Leafs to get one back if they want. Or they could find the very elusive waiver exempt, AHL prospect of value they’ve not found to date. I can hope for that, but I don’t expect it.
I don’t think Babcock will wait out ten starts for Sparks if he keeps playing like he did in his first one. But I don’t think anyone should push the panic button after one game, and it won’t take ten good starts to overcome this setback either. All of the above applies to Jeff Glass on the Marlies as well (six goals against in one start), but it’s much easier to add someone to the AHL roster than it is the NHL.
Just wait. Advice and a threat both. Just wait, and we’ll find out come November if this is a problem or not.