The National Hockey League is expected to expand the statistics shown on their website, to include modern advanced stats, starting February 20.
For those in the analytics community, this marks validation, a sign of official acceptance that has been a long time coming.
Considering the esoteric nature of their current names, though, there is the very real possibility that even as these new stats gain acceptance, they may not necessarily be under the same labels when they are posted on NHL.com. Nor should they be. Go ahead, make the change now.
So, what stats are going to be included?
In the first roll-out, it’s expected that NHL.com will include:
Corsi – total shot attempts
Fenwick – unblocked shot attempts
Zone Starts – the percentage of shifts that a player starts in the offensive zone, relative to shifts started in the defensive zone
PDO – combined on-ice shooting and save percentages during 5-on-5 play
Per-60-minute scoring rate stats
Penalties drawn and taken
Distinguishing between primary and secondary assists
If these don’t appear to be particularly complex ideas, welcome to the NHL’s analytics era which really is not very complicated.
What’s in a name?
Corsi, which is essentially the foundation of modern hockey analytics, was so-named because Tim Barnes (aka Vic Ferrari, currently employed by the Washington Capitals), writing on the Irreverent Oilers Fans blog, liked the look of Jim Corsi, who was the Buffalo Sabres’ goalie coach at the time.
Attribution is responsible for Fenwick and PDO, as they were handles for others in that same stats community. Matt Fenwick was behind the idea of not counting blocked shots and, even more obscurely, Brian King went by the online handle of PDO, and was credited thusly with the concept of combining on-ice shooting and save percentages for a number that is a shorthand measure of luck.
These are interesting back stories for the current names of these stats, but none of those labels remotely indicates what the statistics measure and that has caused some uneasiness among those in the statistical community, especially anyone that has experience in a sport like baseball, which is further along the stats spectrum with an alphabet soup of statistical names, but they are all relatively descriptive of what they are trying to measure. Maybe the acronym WHIP doesn’t mean much on sight, but Walks and Hits per Inning Pitched is a simple enough concept to understand.
I asked my colleague on TSN Hockey Analytics, Travis Yost, what he thought about the idea of new names for hockey's fancy stats.
“Abbreviation is key,” the esteemed Mr. Yost began. "Basically any kind of abbreviation you come up with for shot attempts is clunky and, to be honest, not user-friendly.”
New names for Corsi (shot attempts) and Fenwick (unblocked shot attempts) are easy enough, though Travis is right, short forms (ShA, UShA?) are a little more challenging. As for PDO, would On-Ice Percentages (OI%) suffice? New handles take some getting used to, but there is a large group of hockey fans that aren’t remotely used to the names being used for the stats now.
The issue is one of accessibility. For those of us that have been using advanced stats, Corsi, Fenwick, PDO and so on are already understood, but the important part of those stats isn’t the names, it’s the concepts, the advances in understanding the game, and those are going to live on no matter the name.
Not to be lost in all of this is that the new era of player-tracking chip technology appears to be getting closer on the NHL horizon, likely within the next year or two, and that will bring a whole new bunch of stats to the table, so if these metrics are going to gain some acceptance and understanding, now’s the time.
Yost thinks that the time for changing names has passed. “Six years ago, I would've spearheaded a change,” he said. “Now, I don't see much of an argument.”
While I totally understand Travis’ position on this, I also disagree. I see how difficult it has been to gain traction with the names as-is. It’s one thing for those who are already into advanced stats to think nothing of the quirky nature of the names – once you understand the concepts, the names don’t mean a whole lot – but putting these statistics online is more than just a nod to the small, but growing, percentage of fans that are already versed in advanced stats; it’s about gaining more fans, more eyeballs, more page views, and if the names of advanced stats are even perceived as a hurdle, that will be enough to initiate change.
Remember, this is a league that has had no trouble renaming its divisions, multiple times, getting away from names that were steeped in NHL tradition, so why would there be any hesitation to change names away from online handles?
The modern analytics era (sometimes referred to as the Behind The Net era, based on Gabe Desjardins’ website that started popularizing these metrics) effectively started in 2007, so it’s not ancient history but, for those that bought in relatively early in the process, it has been a constant struggle for acceptance.
Recognition isn’t the goal of using advanced stats but, too often, advanced statistics have been altogether ignored by mainstream media, so recognition is a starting point. Even as sample sizes got larger and years of research showed the predictive value of using these stats, they have been denigrated as the domain of the nerds, not worthy of the attention of “real” hockey fans.
But then NHL teams started hiring statistical analysts – more publicly and in greater numbers than ever before – in the summer of 2014, the watershed “Summer of Analytics” and, while some holdouts might try their best, it is harder to ignore the so-called fancy stats.
Even though these new statistics have been taken straight from NHL play-by-play data – it’s been there all along – there has been nothing official with the league, meaning that fans of such stats had to find their data on hobby sites. Extra Skater rose to great popularity last season, as the preferred domain for fans and journalists alike, but then founder Darryl Metcalf was hired by the Toronto Maple Leafs, leaving more room in that marketplace. In that void, there is still wonderful data being compiled at Puckalytics, War on Ice, Natural Stat Trick, Hockey Reference and others but, in terms of acceptance, that’s not the same as having league-sanctioned data.
To this point, NHL data has been a compilation of the same stats to which hockey fans have become accustomed – goals, assists, plus-minus, hits, faceoffs, etc. – and they aren’t going away, but by making the decision to include newer statistics, the NHL now becomes an official provider of numbers that had previously been the province of hobbyists and *gasp* (hushed tone) bloggers.
By officially accepting these stats on their website, the NHL also opens up to having them used in arbitration. That’s not a huge deal, since very few cases ever make it in front of an arbitrator any more, and explaining more evolved stats to an arbitrator who isn’t necessarily well-versed in their value may pose challenges, but if a player looks better using advanced stats, you can be sure agents will use that as leverage.
Following new statistics might require some getting used to, but it has been that way for everyone at one point. No one was born an advocate of analytics in the NHL; everyone that is already on-board had to have that lightbulb moment at which they were convinced and decided to pay more attention to the new numbers.
The truth is, whether the names are changed or not, the takeaway should be a positive, that the league is moving forward with new data that will provide more useful information for their fans.
If that means more people gain a better understanding and hockey conversations get smarter, then that's a win; about as basic as a stat can get.
Scott Cullen can be reached at firstname.lastname@example.org