The only stat that matters is goals

The only stat that matters is the goals up there on the scoreboard. This is one of those statements that is simultaneously totally true and also totally full of it. It’s the sort of faux wisdom that gets trotted out by people who think looking at hockey through a lens of shots tells you nothing.

At no time will I ever argue with the “analytics is dumb, LOL” crowd.  Organizing my sock drawer is more rewarding, but within the truth part of the put down they are so fond of is something interesting. You do actually have to score goals to win games, and ignoring that and only ever looking at shot differentials is as misleading as never looking at shot differentials and only looking at the goal results.

One of the ways the Leafs successfully tanked for Auston Matthews was by having such a low quality of personnel on the front end that they could execute a system and have decent shot differentials, but they never scored goals. (I’m never talking about shots on goal when I say shots.) Built into the reality that shot differential correlates with future success in wins or goals scored is the reality that very little of the mass of data used to achieve that result is made up of tanking teams full of sub-replacement-level players.

There are other aspects to shot differentials not being the whole story. There’s a thing that happens when the Corsi percentage goes way up and the Shooting percentage goes down.  There’s the current state of parity in the NHL where most teams could on any given night beat all the teams in the league. There’s very few free rides against bad teams, and a lot of teams low in the standings aren’t really all that bad.

The rise of the Expected Goals models is partly an answer to these and other weaknesses in bare shot differential as a descriptor of team strength.  Expected Goals, despite the name, is just Corsi, or shots, weighted for location and type. The models, if done well, give you an idea of what an average team “should” be scoring. But it’s not a benchmark you are shooting for. You want to exceed these expectations, unless your goal is to be merely average because drafting 15th sounds like fun.

To see how well you’ve exceeded the expected, we bring in the actual goals scored. Goals - Expected Goals is how much you are above average. If we look at that for every team, we can judge who has performed well, and who hasn’t.

Enter Cole Anderson. He has an expected goals model that he uses mostly to produce some goalie analysis that is really interesting to me, and we will talk about that sometime soon. His model makes some assumptions about how to weight shots and is not quite the same as other expected goals models.  You can dig into it in the archives at his blog or find the details on GitHub.

He used his model to look at teams and the goals they’ve scored or allowed this season — all of them, not just five-on-five (but no shootout or empty net goals like the NHL standings goal differential shows).  He compared the reality of goals in the season to date with the expected goals for and against from his model, and then he broke it down by four components. You get to see how much of the goal differential was from the goalie, the special teams, five-on-five play and shooting results.

It’s complicated, it’s colourful, and it’s really interesting:

First, read the legend and remember that goal scoring comes from luck and skill.

The horizontal axis is the goal differential, which is why Tampa is there on top and Buffalo is down at the bottom, as the hockey gods have decreed.  The numbers are the total differential.

I’ve been looking at this on and off for a few days as it’s changed a bit as teams play more games, and I’ll tell you what I see here:

• No one, not even Buffalo is wholly bad. They have the tiny shred of special teams on the positive side.
• There are only three teams with no negative component: Tampa, Boston and Toronto. In fact, if you just pumped some air into all of Toronto’s components they’d be Tampa.
• If you remove the red goalie component from Tampa, that is, if you secretly replaced their goalies with league average grey nothings, they’d still lead the league in goal differential.
• I hate Tampa right now.
• If Winnipeg could ever really fix their special teams and get a goalie, they’d be Tampa.
• Boston is leaning very heavily on Tuukka Rask. They’re good, but he is as important to them as Vasilevskiy is not to Tampa.
• Vegas is good and lucky.
• Nashville is shooting skill/luck and their goalies, nothing else.
• LA is the same only more so.
• Toronto is underperforming the “Shooting Lift”. Lack of goals from that area lately are one reason why everyone has decided they’re bad. They aren’t. I’m not sure how good they are yet, either, but they aren’t bad. However, the size of their much more important blue even strength segment is what’s holding them back, just behind Boston.
• Dallas says that goalie deal Yzerman made was pretty smart.
• St. Louis has all the playoff skills going. If they worry about special teams too much, that’s likely a waste of time.
• The Rangers are nothing but Lundqvist.
• The Capitals are Nashville and LA, only more so.
• Chicago is really, really, really nothing but Crawford, who is the only goalie who should even be discussed for Vezina.
• Colorado is a case study in how much of scoring goals over expectation is skill and how much is luck. That’s all they’ve got, so they better hope it’s not going away.
• And I’ll stop now, because the rest of the teams are just a couple of good teams that can’t shoot and bad teams that aren’t interesting, except Anaheim. That line in the chart says: Fire the coach. Everything coachable is bad on a team with a roster miles better than Colorado’s. /

So what do you think? Should we force Tampa into the sea and take back their blue and white stuff because it’s ours? Should we temper our fear of Boston, or redouble it because hot goalies win you cups? And most important, how do you pump air into those Toronto components, particularly the blue one? That has to be the next step in the Leafs development arc.

Is it time to make trades for Toronto, or is it time to wait for them to keep growing from within? Knowing the answer to that is what would get you a job as special advisor to Lou Lamoriello, but like most choices presented in that way, the answer is likely do both.