#FancyStats Breakdown: Corsi vs. Fenwick
Two of the most commonly used analytics in hockey are Corsi and Fenwick. When discussing the use of these two measures, including the differences and similarities between them, it’s important to note that they are often used interchangeably (though that's not a universal acceptance).
The key difference between these two statistics: Corsi measures shots on goal, goals, missed shots and blocked shots, whereas Fenwick does not take blocked shots into account.
In more cases than not, there won’t be an enormous disparity in Corsi and Fenwick percentages but Fenwick is often applied over longer stretches to show a team’s possession trends.
Why use Fenwick and not Corsi? Some in the analytics community feel that shot blocking is driven by luck as opposed to it being skill-based (as has been discussed in the past), so in the case of Fenwick, blocked shots are excluded.
Take, for example, the Blue Jackets’ top three Fenwick for (FF%) players in 2013-14:
James Wisniewski – 54.1%
Nick Foligno – 52.0%
Brandon Dubinsky – 51.8%
Compared to the team’s top three Corsi for (CF%) players:
James Wisniewski – 54.1%
Brandon Dubinsky – 52.2%
Ryan Murray – 51.6
*Foligno is fourth at 51.2%
The differences aren’t too drastic and, in some cases, the Corsi and Fenwick numbers are nearly identical. That’s why many teams who regularly apply analytics in hockey operations track Fenwick on a “rolling” basis, or over the course of 10-15 games to show how their team’s puck possession is trending.
Take the Blue Jackets for example. They monitor both Corsi and Fenwick on a team and individual basis, but when it comes to a metric used to contribute to the discussion on a player, they will most often key on Corsi, according to Josh Flynn, director of hockey administration and analytics lead.
“I generally look at Corsi because it’s tracking more events – blocked and unblocked shots combined,” Flynn told BlueJackets.com. “Over a window of a couple of seasons, Fenwick Close numbers have been predictive of team success but I think Corsi has been just as predictive.
“There usually isn’t a big disparity in the percentages for Corsi and Fenwick. (Fenwick) is a good approximation of possession, as is Corsi, but Corsi counts more events.”
Let’s take a broad look at Fenwick as it relates to the Blue Jackets’ 2013-14 season. As you can see in the charts below, the Jackets were a markedly better possession team as the season progressed.
There are some low points that are easily identifiable (a 7-0 loss to the Oilers in November, for example) and an upward trend that, not surprisingly, coincides with a franchise record eight-game winning streak in January, and a furious push to the Stanley Cup playoffs in March and April.
The first chart, via ExtraSkater.com, shows the Blue Jackets’ cumulative 5-on-5 Fenwick for (FF%) percentage when the score was close. “Close” is defined by prominent analytics website BehindTheNet.ca as “tie games and in the first and second periods of games within two goals.”
Also from Behind The Net: the best "Fenwick Close" score that any team has achieved in the last five seasons is Detroit's 59.4% in 2007-08.
The second chart, which you’ll see below, is a rolling 10-game look at the Blue Jackets’ Fenwick in close situations.
Fenwick paints a rather intriguing (and often times accurate) picture of possession trends and can be predictive of future winning percentages. Has it yet reached the “mainstream” level of use that Corsi has? In some cases, absolutely, but Corsi is currently the leader.
Could that change in the near term? Flynn thinks it could.
“We use this information to help ask the questions, and it’s another viewpoint to use to ask a question; I can come to the table and say ‘I’ve analyzed this guy’s numbers and I’ve watched him play five times in this time frame,’ which gives us a more complete picture of the player than maybe we had before,” Flynn said.