Throughout the season, while covering the Philadelphia Flyers at Broad Street Buzz, we have begun incorporating “fancy stats” into some of our articles. While we provided some basic information on what those stats entailed, I felt it’d be helpful to have a dedicated article explaining and detailing these advanced statistics.
While advanced Hockey statistics are still in their infancy, there is a lot that can be gained by understanding even the most basic metrics like Corsi and Fenwick.
Through this series I hope to accomplish a couple of goals:
- Provide basic definitions and examples to help you folks understand the context in which the metric can help evaluate team and individual performance
- Attempt to help bridge the gap between “eye-testers” and “stat nerds”
- Introduce people to the growing advanced hockey stats community
- Present these new metrics without coming off as some smarmy douche
In the part one of my series here for Broad Street Buzz on hockey’s advanced statistics I went over the Corsi shot metric. You can find that informative article here.
Today I will be introducing you to the Fenwick shot metric.
Where did the name come from?
What is a Fenwick?
Fenwick, as defined by Extra Skater:
Fenwick is the number of unblocked shot attempts by a team or player.
And here is the basic equation one can use to get Fenwick totals individually or at the team level.
Fenwick = (Shots On Goal For + Missed Shots For) – (Shots On Goal Against + Missed Shots Against)
By eliminating blocked shots, Fenwick is able to accomplish two things Corsi does not:
- You eliminate penalizing defensive-minded defenseman that may be skilled shot blockers
- Players that get shots passed an opposing player trying to block it are rewarded with a Fenwick event
What does Fenwick tell us about a team’s performance?
Fenwick is primarily used as a proxy to visualize scoring chances. It has become increasingly more evident that team’s that have strong Fenwick numbers are inherently more likely to be Stanley Cup contenders.
Chris Boyle, writer at the Hab’s blog Eye on the Prize, made a nifty little chart that helps strengthen the argument that strong Fenwick teams are more likely to make the playoffs. While you will find some outliers on that chart, like our own Philadelphia Flyers, you will be able to see the strong correlation.
While Fenwick correlates with time of possession, Corsi in smaller sample sizes (individual games) is actually a slightly better metric for it. The beauty of Fenwick though is its predictive value in regards to a team’s future success.
In my previous article on Corsi, I had mentioned I would use the Toronto Maple Leafs 2013/14 season as an example of how Fenwick can successfully predict future results for a team that is relying on unsustainable luck.
In the chart below from Extra Skater you will see the Toronto Maple Leafs rolling even-strength close FF%.
As you can see, the Toronto Maple Leafs was an abysmal team in terms of their 5v5 close FF%. Even though they carried a record of 32-22-7 in their first 5 months, they crashed in the last two months, going 6-14-1.
Many in the Toronto media were struggling to find answers for their collapse. However, many of the analytic-centric bloggers who cover the Maple Leafs and outside hockey observers saw it coming.
So, how did they see it coming?
Well they observed their close FF% and then took a look at the Maple Leafs PDO.
In my next article on advanced statistics I will go over PDO, so for the sake of brevity, PDO is a metric that evaluates a team’s luck. So a team with a PDO significantly over 100 is considered to have good luck while conversely, a team with a PDO significantly under 100 is considered to have bad luck.
The Toronto Maple Leafs rolling 10-game PDO, while a small sample size, helps illustrate how PDO helped play a part in their collapse. .
As you can see the Maple Leafs were playing with fire. Statistically a team just cannot survive such a low FF differential and a high PDO. Eventually the team will regress back towards the mean (100). So those analytic-centric bloggers and outside observers saw this data and were warning people of their imminent collapse. Few fans and even fewer Toronto beat writers listened, and once the final game was played, the Toronto Maple Leafs found themselves sitting at home once the NHL playoffs started.
So, the biggest take away regarding Fenwick at the team level is its predictive value when it comes to a team’s future performance. However, since there are less data values associated with the calculation of Fenwick, its predictive power is strengthened when dealing with larger sample sizes.
What does Fenwick tell us about a player’s individual performance?
As I stated earlier in this article, Fenwick is a better indicator of scoring chances. While Fenwick correlates with puck possession, Corsi does a slightly better job with it.
Now knowing that, let me present you with an example of how Fenwick data is displayed, tallied, and how to interpret the resulting math.
Fenwick Is Easy
Fenwick Shot Differential
In the chart above you see that Player 1 was on the ice for 12 Fenwick For events and 5 Fenwick Against events. The resulting differential is a +7 for Player 1. Player 2 was on the ice for 14 Fenwick For events and 8 Fenwick Against events. His resulting Fenwick differential was +6.
Since most sites present these differential’s in percentages, I’ll help you with the math.
Fenwick For / (Fenwick For + Fenwick Against) = Fenwick Percentage
Player 1: 12 / (12 + 5) = .706
Player 2: 14/ (14 + 8) = .636
As a side note, these statistics that I used are not artificial. Player 1 is Kimmo Timonen and Player B is Mark Streit. The data is from the Philadelphia Flyers game on 4/6/2014 versus the Buffalo Sabres.
So, now that we have the finished math, what can we interpret from the data?
Similar to how we read Corsi, any resulting percentage that is over 50% can be considered good.
In the finished math you can see Kimmo finished with a FF% of 70.6 and Streit 63.6%. Both very good numbers indicating that both players were on the ice for more scoring chances for than against. However, even though Mark Streit was on the ice for more FF events, he had more FA events. What this should tell you is that even though Kimmo had 2 less FF events, he was the more efficient player while on the ice.
The outro and an advanced hockey statistics term I forgot to mention.
Before I sign off, I want to introduce one more term to your growing advanced hockey statistics vocabulary.
I should have introduced this term in my Corsi article but since the term is also used with Fenwick I’ll present it here.
When navigating to a site like Extra Skater or Behind The Net you will find the term Rel (relative) in conjunction with Fenwick and Corsi stats. The idea of Rel is to help compare an individual’s ability regarding possession (Corsi) and scoring chances (Fenwick) to that of their teammates when that player is not on the ice.
To provide you a quick example of how FF% rel and CF% rel works I’ll use the stats from the same game I used when I talked about how Fenwick can help evaluate individual performance.
What Can Rel Do For You
In the table above you see each player’s CF% and FF% along with their corresponding relative percentage.
As you can see, Kimmo had a +3.9 FF% rel which means that while he was on the ice the Philadelphia Flyers directed about 3 more shots (per 60 mins) at the opposing team’s net than when he wasn’t.
In the case of Andrew MacDonald, even though he carried an impressive FF% of 62.5 his FF% rel was -7.3. So conversely, a negative FF% rel will tell us that when Andrew MacDonald was on the ice the Philadelphia Flyers had about 7 more shots (per 60 mins) directed towards their own net then when he wasn’t.
Where can I go to find these stats and who can I follow to learn more?
Twitter people who are good with stats.
In the next installment of my introduction to advanced hockey stats I will be explaining PDO and why I hate that the acronym has nothing to do with what it actually represents.
Stats Provided by ExtraSkater.