National Hockey League
Analytics go mainstream as NHL.com adopts advanced stats
National Hockey League

Analytics go mainstream as NHL.com adopts advanced stats

Updated Mar. 4, 2020 2:22 p.m. ET

GLENDALE, Ariz. -- Coyotes coach Dave Tippett believes wholeheartedly in the importance and value of hockey analytics. His secret and sizeable stash of advanced statistics allows the coaching staff to analyze game situations down to the tiniest detail. 

Ask Tippett about Corsi and Fenwick, however, and you'll get a scowl that speaks a thousand words.

"I'm not a big fan of Corski or Fenski," said Tippett, who knows those aren't the actual names of the so-called 'possession' stats. "I think those stats are way overrated."

ADVERTISEMENT

Why?

"Because they come off NHL stats," Tippett said. "They're not reliable."

There are infamous disparities between arenas when it comes to hits, turnovers and even shot tracking by the league's stats keepers. Analytics experts argue there are ways to account for those disparities, including large sample sizes. Tippett says there's no point in relying on data he knows is faulty, data he says doesn't necessarily track what he wants to track. 

Despite the disagreement over their value, some of the better known analytics went mainstream Friday when NHL.com added a host of enhanced statistics to the site, including five variations of Corsi and Fenwick -- called shot attempts (SAT) and unblocked shot attempts (USAT) -- that take into account score situations, relativity and percentages.

The site also offers additional stats that include goals and assists (primary and secondary) per 20- and 60-minute averages; shot distance; zone starts; penalties taken and drawn; and PDO, a statistic that's been used to gauge the sustainability of streaks, or puck luck.

This is the first phase of the rollout by the NHL and SAP, the software corporation with which the league has partnered to improve NHL.com's stats pages in the short term, before integrating player tracking data down the road.

COYOTES vs. LIGHTNING

When: 6 p.m. Saturday
Where: Gila River Arena, Glendale
TV: FOX Sports Arizona
Season series: Lightning leads, 1-0
Injuries: Arizona F Mikkel Boedker (splenectomy) is out indefinitely. C Martin Hanzal (back surgery) is likely out for the season. D Zbynek Michalek (upper body) is day to day. F Alex Bolduc (mid-body) could play. Tampa: D Matt Carle (abdomen), D Radko Gudas (knee) are out.
Quick facts: Tampa won in Anaheim on Thursday to climb within a point of Montreal for the Atlantic Division lead (the Canadiens have two games in hand). ... Tampa leads the NHL at 3.22 goals per game and boasts the Eastern Conference's best 5-on-5, goals for/against ratio at 1.29. ... The Lightning is just 14-12-5 on the road. ... C Steven Stamkos leads the team in goals (34) while C Tyler Johnson leads the team in assists (34) and points (56).

"This is definitely a big milestone for analytics," said Michael Schuckers, an associate professor of statistics at St. Lawrence University whose specific work in hockey analytics can be found at statsportsconsulting.com. "I'm excited about this and it's definitely an improvement; a step in the right direction. But as someone who analyzes a lot of data, what we have from the NHL is pretty crude. There are things we can continue to do to improve on these statistics."

One of those improvements could come through the league's plan to insert tracking devices into players' uniforms and the puck to gain a better sense of the myriad things a player does when he's on the ice. 

It may also provide a more accurate read on time of possession than Corsi and Fenwick by actually tracking when one team or player has the puck.

"It's obvious that in most games, if you have more shot attempts, you probably possessed the puck more, so there is a correlation between Corsi and possession," said Coyotes assistant GM Darcy Regier, who is spearheading an increased focus on analytics within the organization. "But those stats don't measure possession. They measure shot attempts. You'd like something a little more solid and tighter than that and the chips in the puck and uniforms might provide that." 

The mainstreaming of analytics began in earnest this summer when a number of teams hired outside experts to produce and analyze data, including Toronto, Florida, Edmonton and New Jersey. 

Many of the remaining members of the analytics movement met at a conference earlier this month in Ottawa to discuss the present and the future of advanced stats. The conference was organized by Schuckers, Shirley Mills of Carleton University, and Rob Vollman, author of Hockey Abstract, whose player usage charts are an invaluable tool for understanding players' roles within the context of games.

Part of that conference was spent celebrating the fact that analytics are finally gaining acceptance within a community where more than a few old-schoolers still believe the eyeball test is more valuable. Vollman understands that sentiment.

"If I walked up to you and said 'I've got something that's going to help you understand hockey and enjoy it more,' you might think 'I already do enjoy it and understand it' so the onus is on me to show what I have can actually improve your understanding and that's not an easy thing to do," Vollman said. "Looking at hockey in terms of numbers is difficult to do and even more difficult to explain, but when we start to predict things and those things actually happen, that's when people sit up and take notice."

"We predicted Toronto would collapse last season and even said why, and when it actually happened, people thought, 'Oh, maybe this there is some merit to this after all.'"

Vollman does not discount the expertise of NHL personnel in evaluating players' based on what they see, along with the intangible aspects of leadership, heart and work ethic. But he believes it is short-sighted to discount the value of analytics. That attitude has permeated some levels of the league and is apparent in Tippett's and Regier's use of advanced stats.

"Analytics aren't about destinies," Vollman said. "They're about probabilities. Even the best will be right about 55 percent of the time, so when you realize that we're sitting at around 55 percent, you incorporate that into your evaluation." 

Regier acknowledged NHL personnel might bring their own biases to the table when analyzing data. He also cautioned the analytics experts bring their own biases; that's it's simply human nature. But those biases haven't stopped Regier from having conversations with a number of outsiders, including Vollman. 

"We're trying to get a little closer to the front lines in order to get a better understanding of what's out there, what might be coming, what the league is doing and what we can do with that information," he said. "Some of the biggest advances in the industry have come from outside the league so it's good to talk to outsiders and it's great that these analytics guy have gotten involved. 

"It's great for fans to have this additional information available and it's great for the conversation as we try to gain a better understanding of the game."

Follow Craig Morgan on Twitter

share


Get more from National Hockey League Follow your favorites to get information about games, news and more