Advanced stats are no longer just for the nerds who follow hockey and blog from their moms’ basements.
“It has become more a part of the mainstream conversation,” said Frank Provenzano, who served as assistant general manager for the Dallas Stars from 2006-13. “You’ll see it now in mainstream media when they are talking about teams and on broadcasts.”
@ngreenberg I’m watching the Sabres broadcast on Game Center live and they just displayed corsi numbers from the first period. Pretty cool
— Adam Stringham (@Stringhama) December 14, 2013
— Derek Jedamski (@TheHosers_DSJ) December 14, 2013
— DJC (@oilfaninvan) December 19, 2013
You can also find it in more and more front offices across the NHL.
“I think we are obviously behind baseball as a sport, we’re behind basketball as a sport, but I think we are right there,” said Don Fishman, assistant general manager for the Washington Capitals. “I think hockey teams are increasingly using statistics, decision-making methodology, analytics, advanced statistics and cameras in their decision making, and it is increasingly growing as a sport.”
Aside from the traditional statistics such as goals and assists, advanced statistics offer greater insight into which players are performing well in crucial areas of the game, including puck possession, defensive play and situational ability. That insight helps create a better valuation for a player.
“You might not get advanced statistics in negotiations, maybe two out of every 10, but in most of them it is definitely useful to evaluate on your own side how you value the player,” Fishman explained. “You look at advanced statistics for how you value the player because you want to go into the negotiation for an unrestricted or restricted free agent with a full and complete picture, so you want to look at all the statistics available.”
That is especially useful for players who typically do not put up the more tangible boxcar stats, such as goals and assists.
“Defensive defenseman are unique because you won’t get huge point totals for them,” Fishman said, “but you have time on ice, which is the generally accepted statistic, and you go to things like goals for, goals against and plus/minus. But now you have the advanced statistics that a lot of people use about shot attempts and more advanced ways of looking at plus/minus because it is now generally accepted that plus/minus is not a very reliable stat, so analytics and statistics lead to a better way of looking at defensive defensemen and valuing them.”
Time on ice, readily available on NHL.com, is not an advanced stat, but it makes sense that better defenders would play more of the game. But Fishman says you can’t look at any raw number in a vacuum.
“Situational play — power play, even strength and penalty kill — you want to look at time on ice across all three situations,” Fishman said. “You want to look at points across all three situations. It’s really three types of very different play. It’s imperative to look at them differently.”
A good example of this is Toronto blueliner and upcoming restricted free agent Cody Franson. He plays 21:18 per night including 3:11 on the power play. Of his 18 points this season, 13 have been tallied with the man advantage. However, ignoring special teams and lead-protecting situations, the Maple Leafs have been out-Corsi’d 307 to 403 with Franson on the ice.
The shot attempt stats, such as Corsi — which was first used by Buffalo Sabres goaltending coach Jim Corsi to measure the quality of shots his netminders were facing — and Fenwick measure how often players or a team are in possession of the puck by tallying every shot (including those that miss the net or are blocked) directed at net while they’re on the ice. But Frank Provenzano sees a problem preventing the stats from becoming widely accepted: “They [have] really dumb names.”
“They just aren’t intuitive,” Provenzano added. “Coaches are already doing a lot of these similar analysis; they just aren’t calling it Corsi or Fenwick. Every coach I know wants to play a puck possession game, but really what is Corsi or Fenwick? It means you have the puck and are trying to get it on net. It isn’t all that advanced of a stat, but it is a good stat.”
Fishman, however, doesn’t think the “dumb” names are a dealbreaker. “I think a lot of people get pigeonholed into thinking analytics means advanced math and advanced statistics and funky names, and to me analytics means so much more than that. It is allowing statistics and decision-making and situational analysis to improve your thinking and understanding of the players and what works best: when should you pull the goalie, who should you use in the shootout and which goalie should you use against a particular opponent.”
Teams are craving that type of analysis, and while the Chicago Blackhawks and Los Angeles Kings make no secret they have analytic-types on staff, there are other NHL teams who turn to organizations like the Sports Analytics Institute, co-founded by Mike Boyle and Kevin Mongeonto, to provide that type of information.
“We have models Kevin has developed that help us break down the game beyond the traditional statistics,” Boyle said. “We can tell how a player is really performing when you take into consideration the quality of his linemates, opponents, score and state of the game instead of raw numbers. Is he padding his stats when the game is out of reach or is he delivering when it matters? Is he helping his linemates or is he benefiting from the quality of his linemates? So we do a regression-type of analysis to go beyond the straight statistics to see what is really happening out there.”
If the data is available and the geeks, coaches and execs agree on their importance, it is only a matter of time before the Moneypuck revolution sweeps into every NHL front office, right? Not so fast.
“What some of the analytics people fail to recognize, or historically have failed to recognize, is that they don’t factor in the actual dynamic and accountability of making a player decision because there are consequences of being wrong,” Provenzano said. “Our system is very imperfect, but it is like anything else: People accept the errors of conventionality more than unconventionality. If you do something different and are not right everyone is ready to pounce on your error. It will eventually become accepted, just not as fast as a lot of analytics people want.”