O’Donnell: What’s up with Corsi this season?

Shot attempt metrics have become widely accepted around the blogosphere because of the reliability and predictive power they carry. Sample size for shot attempts grows much faster than sample size for goals, and as a result, metrics such as Corsi and Fenwick are more reliable assessments of a team’s true talent than goal metrics, especially over a small sample of games.

We’ve known this for quite some time now, and have seen teams that experience short-term success without strong shot attempt numbers fall back to Earth over long periods of time. Prominent examples include the 2012-2013 Toronto Maple Leafs, the 2013-2014 Colorado Avalanche, and 2014-2015 Calgary Flames. All three experienced regular season success, and the Flames even won their first round series against the Vancouver Canucks, but all three also immediately fell down the standings in the next season, due to the fact that their impressive performances in the prior year were largely the result of random variance; some high percentages (PDO) made the teams look better than they actually were.

The true predictive power of shot attempt metrics lies in predicting future Goals For percentage at 5 on 5, but over the past decade or so, Corsi For percentage has been a pretty reliable indicator of standings points. From 2007-2015, the R2 between CF% and standings points is .4, which isn’t exactly incredibly high, but is still noteworthy.

shane corsi 1


The 2015-2016 season has been a very strange year for shot attempt numbers, howerver, as several teams have gone on ridiculous PDO binges despite getting outshot at 5 on 5. Here’s a list of all the teams with a CF% below 49.5%, and a PDO over 101.0.


Team CF% PDO
New York Rangers 48.2% 102.7
Florida Panthers 48.1% 102.4
Minnesota Wild 49.1% 101.6
Arizona Coyotes 47.3% 101.4
Ottawa Senators 46.1% 101.3
Colorado Avalanche 44.6% 101.1


Usually we see one or two teams really ride the percentages up the league standings. This year, there have been six.

Conversely, there are plenty of teams who have dominated the opposition at 5 on 5, but haven’t gotten the bounces. Here’s a list of teams with a CF% above 51.0% and a PDO below 99.5.


Team CF% PDO
Anaheim Ducks 53.3% 96.9
Carolina Hurricanes 53.2% 97.5
Nashville Predators 54.2% 98.8
Chicago Blackhawks 52.1% 99.2
St. Louis Blues 53.0% 99.3


As a result, the relationship between CF% and standings points (or the number of points a team is on pace to finish with) is much lower; the R2 between the two in 2015-2016 is only .18.


shane corsi 2

So what exactly is going on with Corsi this season? For starters, a couple of teams can be expected to outperform their 5 on 5 shot metrics this season. The Washington Capitals are one example, as they have an elite net minder in Braden Holtby (who leads all goaltenders who have played 30 games with an insane .931 SV% in all situations), and the league’s most dangerous power play, led by Alex Ovechkin, Evgeny Kuznetsov, and Nicklas Backstrom. With strong goaltending, and a positive special teams differential, the Capitals should pick up more victories than their 5 on 5 numbers would suggest; sure enough, the team is 2nd in the league standings, but 12th with a 51.4% CF%.

The New York Rangers are also an example, as they have an elite goaltender in Henrik Lundqvist, and the Boston Bruins are another example, as they (like the Capitals) have an elite net minder in Tuukka Rask, and an incredibly efficient power play.

There are also teams that can be expected to perform at a level well below their shot attempt numbers. The Carolina Hurricanes are 6th in the league when it comes to CF%, but have very little offensive shooting talent (behind Jeff Skinner and Eric Staal, the team is severely lacking), and one of the league’s worst goaltenders (Cam Ward has a career .909 SV%).

Other teams, such as the Winnipeg Jets, are suffering because of special teams. The Jets are a solid 5 on 5 team, but take a ton of penalties (they’ve been shorthanded more than any other team in the league), don’t kill off a high percentage of those penalties (27th ranked PK), and fail to score on a good number of their own power plays (29th ranked PP).

That doesn’t mean that every team can be expected to out-perform (or under-perform) when compared to their shot attempt metrics. The Florida Panthers do have a good goaltender in Roberto Luongo, but this season, Luongo has been uncharacteristically good; the last time Bobby Lu posted an all situations save percentage over .930 was in 2003-2004.  It would be unreasonable to expect the Panthers to continue to have a PDO over 102 for the rest of the season, simply because 102 is rather high, especially for a team that doesn’t carry elite shooting talent.

The Arizona Coyotes and Anaheim Ducks are two teams that have performed much differently than one would expect based on shot attempt metrics, and it’s mainly because of PDO swings. The Ducks are a strong CF% team, but sit near the bottom of the standings because of an absurdly low 96.9 PDO. The Coyotes, on the other hand, are one of the league’s weaker possession teams, but have ridden a 101.4 PDO to second place in the Pacific Division.

Some of the teams mentioned before, such as the Rangers and the Capitals, have gotten rather lucky as well. Both teams have PDOs over 102, which usually isn’t sustainable over the course of a full season. The Capitals have solid possession numbers, though; the Rangers do not.

Given what we know about PDO (it regresses towards the mean over a long period of time), it would be safe to assume that plenty of the strange cases we’ve seen this season will experience regression, and teams will start to fall more in line with their shot attempt metrics.

Overall, the 2015-2016 season has just been strange, and we’ve seen a very sharp decline in the relationship between CF% and points in the standings. Some of this may have to do with the fact that there are teams that can be expected to overpeform/underperform in relation to their 5 on 5 numbers, but a lot of it has to do with the fact that there have been odd PDO swings for teams. It will be interesting to revisit this when the season ends, and we’ve seen some regression from the teams on the more extreme end of the spectrum.

Perhaps the biggest takeaway from this is that the NHL standings are probably going to change over the second half of the season, and the percentages start to normalize for teams such as Anaheim and Florida. It’s going to be a fun, fun ride.


2 thoughts on “O’Donnell: What’s up with Corsi this season?

  1. One suggestion in regards to the regression: run it on Regulation/Overtime wins rather than points, since the shootout is really an exogenous event that has no relation to overall team play but will still effect points.

Leave a Reply

Your email address will not be published. Required fields are marked *