Years from now, we might look back on the offseason of 2014 as the moment the stats nerds finally took over the National Hockey League.
Over the summer, teams around the best hockey league in the world fought not only over the crop of free agent players available, but also over a bunch of mathematicians and statisticians with zero NHL front office experience. These new hires are bringing the kind of advanced statistical analysis to NHL front offices that some Major League Baseball teams have already been using for 15 years.
You know that scene in Moneyball where Brad Pitt tells a scout to "adapt or die" in the course of a discussion about the value of advanced statistics in baseball? That same scene (perhaps in less-dramatic fashion) is now playing itself out in NHL front offices.
Just as formerly exotic stats such as Batting Average on Balls in Play and Fielding Independent Pitching have changed the way baseball players are evaluated, stats like Corsi and Fenwick are shaking up accepted hockey truths.
Those two stats—Corsi and Fenwick—are the most fundamental and easiest to understand of the new hockey numbers. Both measure puck possession by counting not only shots on goal, but also those that miss the net (and, in the case of Corsi, shots that are blocked). Both stats also have more advanced variants that attempt to isolate the impact of individual players. Zone starts are another common advanced stat, measuring whether a player starts most of his shifts in the offensive or defensive zone.
Traditional ideas like "heart," "sand paper," and hitting—the sorts of things famous hockey pundit Don Cherry throws around—are falling by the wayside in favour of softer statistics like possession and shot attempts.
In other words, some of the barbarity hockey is known for is proving to be statistically weaker than stats that make hockey a game of finesse.
More complicated statistics already exist and the teams that have hired their own advanced statisticians are no doubt developing new, proprietary ones. The big questions now are: How is the new statistical analysis affecting the game we see on the ice, and where do the stats go from here?
Years from now, we might look back on the offseason of 2014 as the moment the nerds finally took over the National Hockey League
In terms of on-ice strategies, teams are already recognizing the value of holding onto the puck. Kent Wilson, whose analysis has appeared in places like the Sporting News and Hockey Prospectus, said he believes the way NHL teams value players will change completely.
"In the very near future, I expect possession players to become more valuable on the market," said Wilson suggesting a guy who keeps control of the puck might be more valuable than a guy who pots a couple goals every few games. "While the best two-way skaters have always been highly valued, I think teams will be able to identify the middle tier guys who drive play but don't stand out quite as much as the elite players."
The statistical revolution has also demonstrated that some numbers aren't as important as we thought they were. Benjamin Wendorf, who writes for The Hockey News and the Chicago Tribune, provided the example of a player who blocks a lot of shots. In the past, he would be congratulated for sacrificing his body for the team, but will that be the case in the future? "Have we even determined if that's a positive or not?" asks Wendorf. "In recent years, we've found that if you block a lot of shots, usually that means that you're playing poor defense, that the other team has the puck."
Even that can be misleading, though. Just because a particular player blocks a lot of shots doesn't mean he is necessarily a poor defensive player; maybe he just plays on a really bad team.
That is one area where the current generation of advanced stats are still limited—separating individual from team performances. "When the NHL fully deploys player-and puck-tracking, we will be able to know where the puck and every player is, at every second of the game," said David Johnson, owner of the sites HockeyAnalysis.com and Puckalytics.com. "We'll be able to look at things like passing, who is carrying the puck up the ice and who is turning over the puck. We'll also be able to determine who are the fastest players, who accelerates the quickest, who is the most agile."
As more information becomes available, teams will place a premium on people with experience manipulating large data sets and extracting meaningful intelligence from them. That's why there was such a rush this year to lock-down many of the best analysts currently working with the new stats.
Still, not everyone is buying in. Los Angeles defenseman Drew Doughty—ironically a darling of the Corsi crowd—recently told his team's blog, "I think that Corsi thing is a bunch of crap." Soon, though, players may not have a choice. When general managers and agents start bringing advanced statistics to the bargaining table, you can bet players will care about their Corsi rating.
With or without advanced stats, though, Doughty is a star. Where the new analysis can really make a difference is in identifying overlooked players, as well as those who may be overvalued. "I'd say a competent analytics department could pay for itself 10 times over if its only mandate was to help the team avoid making terrible bets in free agency," said Wilson. "Being less wrong is the low-hanging fruit of the new NHL stats movement."
The game is on the cusp of a new era and the teams that have embraced it first will have an advantage over the teams that have not. Eventually, the nerds will be proven right and everyone in the league will have to accept it, whether they want to or not. Adapt or die.