Off Season Narrative: The Rise of Fancy Stats?Posted on May 06, 2014 by Adam Stringham
Narrative: Advanced statistics are making their way into the NHL, and the mainstream media is using them incorrectly.
Articles of Note: This piece from Mark Lazerus Advanced Analytics are the Blackhawks' Secret Formula for Success
Second article; Anger towards the media and "advanced stats" from Matt Rudnitsky
My Take: Well the first article by Mark Lazerus is a great look at how good NHL teams are utilizing their own internal statistics to make smart choices. I hope the Capitals' next general manager is part of the "new school" of thought when it comes to team development. I highly recommend that everyone check it out. I do not really have a lot else to say about the first article, I just thought it should be shared.
The second article is a piece explaining how "advanced stats" are not that advanced (true) and how the media is not utilizing these statistics correctly.
For the majority of this year I have felt like the "traditional media" and the non-traditional blogging community were nicely separated. The mainstream media would provide narrative while the blogs would provide stats. It was not until the beginning of the playoffs that I started seeing mainstream media members cite "advanced stats", while the blogging community has been using Fenwick, Corsi, PDO (and more) for years.
Rudnitsky appears to have a big problem with how the media has been trying to use these stats to express opinions on the outcomes of certain series; specifically the Kings-Ducks match up. He cites Sean McIndoe, Pierre LeBrun, Jonathan Willis, and Dave Lozo. This is where I start to have an issue with his article. Mainly because I do not think the media is that far off in their way they used the stats.
"National media loves citing possession stats and acting like they measure team quality" -Matt Rudnitsky
More than just the national media cite possession statistics as a measure of team quality, probably because they do measure team quality (from @BSH_EricT).
Matt goes on to argue that PDO is not an indicator of luck but rather an indicator of skill. While there is some evidence to support that a mildly elevated PDO can be sustained, there is more evidence to show that shooting percentages are generally determined more by luck than by skill. I compiled this graph on PDO and GF/GA.
This graph does help illustrate Matt's point that PDO does not have a huge bearing one way or another on the likelihood of a team going the distance, but anyone who truly likes statistics does not use just one singular statistic and that includes the media. For example: if ANH had a high FF% and a high PDO then individuals would be less concerned about the potential for a PDO regression to be the deciding factor in the series. Anaheim had a shooting percentage in 13-14 that was a full percent higher than their previous season.
Due to the solid goal tending provided to the Kings by Johnathon Quick the Ducks have a shooting percentage of 5.3% and a PDO of 93.3 in their two games against the Kings. It is very hard to win with a PDO of 93.3, just ask Tampa (91.7 PDO) .
Matt selected a quote to use in his article that, to me, indicates that he has failed to grasp the point of Fenwick and Corsi.
"shot differential (which correlates strongly with Fenwick Close) was only predictive of 54.7% of game results".
First off: FF% close correlates strongly to shot differential? That is not really breaking news but I would imagine FF% would be more similar to shot differential than FF% Close.
Second: Corsi and Fenwick are used as indicators of future success, not as a predictor for the outcome of an individual game.
1. I hate the Penguins but cheers to Marc-Andre Fluery for saving his career.
2. The Ducks are in trouble, poor PDO seems to be a consistent barrier for Bruce Boudreau and whichever team he is coaching.
Follow me on twitter @Stringhama and @TalkTheRed