Photo Credit: Kim Klement-USA TODAY Sports

Auston Matthews: Elite Offensively; Average Shooter. How?

Recently, a conversation was sparked by one of the sharpest minds in the public sphere of hockey analytics, DTMAboutHeart. You may know DTM from his Expected Goals (xG) model that is very useful, as well as his GAR model that is very… talked about.

Yesterday, he posted a simple ranking of a statistic he created called “shooting talent”. The purpose of the metric was to isolate goal scoring from being in a good position and goal scoring from being a good shooter. The idea being, if you’re shooting a lot from good areas, that explains a certain portion of your goal scoring. The other portion comes from how good the shot is once you’re in that area. This was the top 30:

You may notice, as many others did, that Toronto’s superstar Auston Matthews is not present.


First, let’s explain a little better how “Shooting Talent” is defined here. The way it’s calculated (generally, I don’t have access to the formula or anything) but this would be the general methodology. First, you establish what the likely shooting percentage for a shot from that location is. There’s a ton of data on shot locations that can be used to study and establish this. Then, you examine what a particular player’s shooting percentage is versus the typical shooting percentages we established. If they score more than you’d expect, that’s an above average shooter.

What we’re talking about is the difference in Expected Goals (xG) versus actual goals. You could call it dxG, like Steve Burtch did with dCorsi. You figure out what to expect, you see if the player did better or worse than that. Once you get to the Expected value, the math is just a simple subtraction. It’s easy enough to follow.

Also, I should note, that the later versions of this graph were called “Finishing Talent”, which may be a more apt descriptor as it highlights that we’re talking about the shot that finishes the play, not all the talent that went into creating that shot.


The title of this post basically sums up the point. Auston Matthews is incredibly good at getting to good scoring areas, which is a big reason why he scored a heck of a lot of goals last season. 40 to be exact! How is it that a player who scored 40 goals in a season is not ranked highly in “Shooting Talent”?

That’s because he only scored a little bit more than we would have expected him to (again, based on shot locations). He got himself in really dangerous areas using his offensive abilities, but this “Shooting Talent” metric shows that the number of times he scored from there weren’t much more than we’d expect. Take a look at the graph for the Leafs:

Look at how high Polak is! The very obvious reason for that is he took so few shots, that when more of them go in than we expect, it can be chalked up to small samples. That’s why the “75% confidence interval” is misleading. We don’t have that level of confidence with each player. It’s important to take these things with many, many grains of salt.

The important point is this: it makes some sense that Matthews isn’t in the top 30 for this particular metric. It’s also not a bad thing that he’s not in the top 30. This isn’t saying Matthews isn’t one of the top 30 goal scorers in the league. It’s just that he creates his goal scoring by being intelligent, by pushing play to the dangerous areas, and by utilizing his excellent skill set to get him there far more often than most. It’s just that he doesn’t usually get the goals by making elite shots.


It wouldn’t be unreasonable to suggest that being a 40 goal scorer who isn’t at the top of this list is a good thing. It means you bring more to the game than just a good shot. Even if this is the way the list looks for the next 10 years, a team will probably still extract more value out of a scoring talent like Auston Matthews than one like Patrik Laine (yes somehow I’m still hung up on this).

The stat really should only hold weight as a descriptive identifier of pure shot talent. Yes, offense is more than that, and we have stats that can quantify that. But as we move towards single-metrics, it’s important that we’re also able to describe where those single metrics come from, and this is something that can be used for that. Auston Matthews is a gem and even though this stat may look like it’s suggesting he isn’t, I promise: he still is.