Better Know a Statistic: Corsi & Fenwick

In this series, we will look at advanced hockey statistics and try to answer two questions: What is it?, and Why should I care about it?  I will try to keep it simple and provide examples to illustrate.

We start with those omnipresent über-stats that fanboys, bloggers, and front offices all drool over, because they are believed to be x-ray goggles that can peer into a team’s or player’s true greatness: Corsi and Fenwick. The fightin’ Fenwick! Is all the hoopla over these statistics justified? Let’s take a look.

What’s with the weird names?

Before we begin, let’s deal with this nonsense. With some every single traditional hockey stat, the name tells you exactly what you’re dealing with, like “shots on goal,” “penalties in minutes,” and “goals.” But not with Corsi and Fenwick. No, no. They are named after people, in order to confuse and frustrate fans. Unfortunately, it’s too late to try to change them, so we just have to learn the stupid names.

Jim Corsi. The man, the coach, the statistic.
Jim Corsi: The man, the myth, the moustache.

But, if you care, Corsi is named after Jim Corsi, currently a goaltending coach for the St. Louis Blues. Some time ago, he was trying to come up with a way to measure how involved goalies are in a game, and he came up with something. Then another guy took what he did and created the statistic we know today, generously naming it after Mr. Corsi.

Fenwick is named after Matt Fenwick, a blogger for the Calgary Flames. He apparently thought it would be neat to modify the Corsi statistic in a very minor way, and of course it ended up bearing his name. Mr. Fenwick can be found on twitter.

What do Corsi and Fenwick mean?

Corsi = All Shot Attempts

Fenwick = All Unblocked Shot Attempts

If you take one thing away from this article, just remember: Corsi = All shot attempts; Fenwick = All unblocked shot attempts.

Corsi and Fenwick are meant to be indicators of “possession,” or how much a team controls the puck in the offensive zone during a game. So Corsi looks at everything thrown at the net; we don’t care what happens to the puck after a shot is attempted. So whether it goes in the net, hits the goalie, hits a post, goes wide, or gets blocked before it gets there, it counts. Fenwick is the same, you just exclude the blocked shots.

Why exclude blocked shots? Supposedly because blocking a shot involves skill or something? Honestly, I don’t know. The rationales I come across kinda suck. Fenwick gets less love than Corsi, probably for good reason.

Examples: Say the Pittsburgh Penguins play the Washington Capitals. At the end of the game, the Penguins took 28 shots on goal, 1 hit the post (not technically a shot on goal), 4 missed the net entirely, and 10 were blocked. Add it all up (28 + 1 + 4 + 10), and the Corsi is 43, meaning Pittsburgh took a total of 43 shot attempts in the game. The Fenwick is 33 (after subtracting 10 blocked shots), meaning the Penguins took a total of 33 shot attempts that did not get blocked by the Capitals.

what are Corsi stats
You can feel the love. (Tom Turk/THW)

The Caps, on the other hand, took 21 shots on goal, 6 went wide, and 6 were blocked. Again, just add it all up, and the Corsi is 33. The Fenwick is 27.

From the Penguins’ perspective, you would say the Corsi “For” is 43, and the Corsi “Against” is 33. If you were talking about the Capitals, you would just switch those numbers around.

Why are they called “advanced” statistics?

To make us bloggers feel smart? I don’t know. Most hockey stats are just about adding things up as they happen (shots, goals, assists, penalty minutes, etc.). Corsi and Fenwick are just like this! You do not need an advanced degree in statistics or math to understand or calculate them.

So calling these stats “advanced” is a huge misnomer. These are simple stats (simplistic statistics?). The only difference with traditional stats is that not many people have paid attention to these until now. Something like “non-traditional” stats would be much more accurate.

Aren’t Corsi and Fenwick also applied to players?

Yes. You can calculate the teams’ Corsi or Fenwick scores just when a particular player was on the ice. A player’s on-ice Corsi or Fenwick is supposed to glean into that individual player’s contribution to his team’s possession.

Example: In our hypothetical Penguins/Caps game, when Sidney Crosby was on the ice, his team took 10 shots on goal, 2 went wide, and 4 were blocked. But the Caps, when Crosby was on the ice, took just 4 shots on goal, 2 went wide, and 1 was blocked.

You would say that Crosby’s on-ice Corsi For (what his team did) in the game was 16 (10 + 2 + 4). His on-ice Fenwick For was 12 (after subtracting 4 blocked shots). His on-ice Corsi Against (what the other team did) was 7. His on-ice Fenwick Against was 6.

In other words, when Crosby was on the ice, his team attempted 16 total shots (or 12 unblocked shots), while the other team attempted just 7 total shots (or 6 unblocked shots).

Aren’t they usually reported as a percentage?

Yes. The basic stats can be manipulated and reported in different ways. Here are two of the most common.

Corsi Percentage = (Team A / (Team A + Team B)) * 100

This tells you the percentage of all shot attempts taken by one or the other team. You can do the same with Fenwick and unblocked shot attempts.

Corsi Differential = Team A minus Team B

This is basically plus-minus for shot attempts.

Examples: With Pittsburgh’s Corsi at 43 and the Caps at 33, Pittsburgh’s differential is +10. Washington’s is -10. This means, obviously, the Penguins took 10 more total shot attempts than the Capitals.

    "I am totally owning this hypothetical example, Geno." .(Geoff Burke-USA TODAY Sports)
“I am totally owning this hypothetical example, Geno.” “Must be the regular season.” (Geoff Burke-USA TODAY Sports)

Pittsburgh’s Corsi% For is 56.58 percent [(43 / 76) * 100], meaning the Penguins took 56.58% percent of all shot attempts in the game. Pittsburgh’s Corsi% Against is what’s left, or 43.42 percent, meaning the Capitals took just 43.42 percent of all shot attempts in the game.

Similarly, Crosby’s on-ice Corsi% For is 69.57 percent [(16 /23) * 100], meaning when Crosby was on the ice, his team took nearly 70 percent of all shot attempts.

What about Corsi or Fenwick in particular game situations?

Yep, they do that to. For example, you can look at either of the stats just when the score is tied (Corsi Tied), or when the score is “close” (definitions I have seen vary, but generally tied or within 1 goal), called Corsi Close.

It’s also common to look just at 5-on-5 hockey. Obviously, when one team has a power-play, it will usually dominate play in the offensive zone, and the vast majority of shot attempts will be in its favor. That doesn’t tell us much about which team is really better at controlling the puck. So if you ask me, we should only ever be looking at even strength when it comes to these stats.

What about some more real-life examples?

Sure. Let’s look at the young 2014-15 season.

So far, the NHL team with the best Corsi% For at 5-on-5 even strength is Chicago, at 57.00 percent. That means the Blackhawks have taken 57 percent of all shot attempts in their games. That is puck control, baby. The team with the worst Corsi% For at 5-on-5 is Buffalo, at 36.09 percent. That means the toothless Sabres have taken barely more than one-third of all shot attempts in their games. They stink.

"Actually, coach, maybe it's better for the team if I sit this one out." Cody McCormick. (Kevin Hoffman-US PRESSWIRE)
Shame, shame, know your name, Cody McCormick. (Kevin Hoffman-US PRESSWIRE)

Likewise, the NHL player (with at least 10 games played) with the best on-ice Corsi% For at 5-on-5 is the great Tomas Jurco (who?), of the Detroit Red Wings, at 63.75 percent. That means, when Tomas the Terrible is on the ice, the Red Wings attempt nearly 64 percent of all shots. Nicely done. The player with the worst on-ice Corsi% For is Cody McCormick, a center for the (wait for it) Buffalo Sabres, at an astounding 31.55 percent. That’s pretty awful.

Hopefully, now, we have a handle on these statistics.

Why should I care about Corsi or Fenwick?

You should care because there is a good theory they matter, but you should be skeptical because there’s no good proof they actually do.

The theory is that better teams possess the puck more in the offensive zone, offensive zone possessions lead to more shot attempts, more shot attempts lead to more goals, and more goals leads to winning. Plus, when your team has the puck in the offensive zone, by definition, the other team doesn’t. It makes sense.

But the proof? That Corsi or Fenwick actually affect things like goals and winning? There is none. To be fair, some attempts have been made to verify this statistically. But these analyses are no good. They fail to account for so many other factors that could be affecting outcomes, things like face-off percentage, hits, home-ice advantage, playing in back-to-back games, etc.

For all we know at this point, Corsi and Fenwick might have nothing to do with scoring goals or winning when you factor in everything else. Or they might be no better (or even worse) than just looking at traditional shots on goal. Or their effects might be tiny in comparison. Or they might matter only in particular circumstances. So all the hoopla is kinda premature at this point.

How would we get proof?

You would have to get a bunch of data, and run some (actually) advanced statistics on it, accounting for the other factors. And, you know what, I’m going to do just that. I will figure out exactly how much Corsi and Fenwick matter, after accounting for everything else I can think of under the sun.

And the results will be sure to BLOW. YOUR. MINDS.

What do you think you’ll find?

My secret desire is that Corsi and Fenwick won’t count for jack-squat, throwing a major wrench in the wheels of the statistical revolution. Because that would be funny. Or I might completely legitimize the whole thing. In the end, my guess is probably somewhere in between. Corsi and Fenwick will have an effect, but it won’t be that huge, and other factors will matter more.

Stay tuned for further research  . . .

1 thought on “Better Know a Statistic: Corsi & Fenwick”

  1. The funny thing about both of these metrics is that they’re composite indices and they suffer the problem that all composite measures do. There can be one component that either deflates or inflates the outcome, and you’ll have no idea which it is. Also, while they may have predictive abilities to tell you whether a player may be successful (mostly retrospectively), they have no bearing on team strategy. Personally I feel like the ‘analytics’ movement shouldnt be about concocting these complex metrics but rather using statistics/data to inform strategy…. I like to call this “translational analytics” – i.e. a GM can use the analytics to draft, a coach can use the data to run practices, set lines, develop strategy.

Comments are closed.