Carolina Hurricanes

Using analytics to evaluate NHL coaching performance

The Canes head coach Bill Peters watches the action during the first period of an NHL game played between the Carolina Hurricanes and the Calgary Flames at PNC Arena in Raleigh, N.C. on Nov. 10, 2014. The Canes beat the Flames 4-1.
The Canes head coach Bill Peters watches the action during the first period of an NHL game played between the Carolina Hurricanes and the Calgary Flames at PNC Arena in Raleigh, N.C. on Nov. 10, 2014. The Canes beat the Flames 4-1.

Over the past year there’s been a gradual acceptance of analytics in professional hockey circles.

A discipline that started as an online argument among fans has gone from the basements of long-suffering hockey markets like Toronto and Edmonton to being routinely discussed in NHL front offices and TV studios.

The basic premise underlying any statistical analysis of hockey is that a player’s performance can be objectively measured – that numbers can tell us things that our eyes miss.

But what if we take a step back and look behind the bench? How do we measure the value of a coach?

Much like a politician who gets credit or abuse depending on economic forces over which he or she has no control, a coach succeeds or fails depending on his players’ hot streaks and slumps.

A mid-season coaching change may be as much about a top scorer’s run of bad luck as it is about the guy tapping players on the shoulder.

Off-season coaching changes present a different problem because they are usually accompanied by roster moves. We’ll never have a perfect laboratory experiment in which coaches play musical chairs and rotate teams over the course of a season, but analytics can still help us discern their impact on a team’s performance.

The accompanying graphic shows how the six teams that made coaching changes at the end of last season are doing based on the following measures: (i) points earned as a percentage of total points available (Point %); (ii) 5-on-5 goals for percentage (GF%); (iii) 5-on-5 Corsi for per 60 minutes (CF/60); (iv) 5-on-5 Corsi against per 60 minutes (CA/60); and (v) 5-on-5 Corsi for percentage (CF%).

To begin with, Hurricanes coach Bill Peters must be scratching his head and wondering what to do next. Brought in to fix a team that was 15th in the league in CF% (50.3) but 22nd in GF% (47.6), Peters was probably the beneficiary of some bad luck on the part of his predecessor Kirk Muller.

Despite getting solid goaltending (5-on-5 Sv% of .926), the Hurricanes just couldn’t find the back of the net last year, posting a 5-on-5 team shooting percentage of 6.82, good for 27th in the league.

Surely all Peters would have to do is stay the course and things would improve.

In fact, Peters has done better, decreasing his team’s CA/60 by 4.9 while keeping its CF/60 fairly constant (-0.5). In other words, for every 60 minutes of 5-on-5 play, Peters’ team is giving up 4.9 fewer shot attempts and generating about the same number for themselves.

Instead of getting about 50.6 percent of the available points last season, the Hurricanes have managed a paltry 32.8 percent this season and sit 29th in the standings.

By no means are the Hurricanes a great team, but they’re not as awful as they look and shouldn’t be front-runners in the Connor McDavid/Jack Eichel draft sweepstakes.

Meanwhile, Barry Trotz left Nashville, took over in Washington, and left his old job for former Hurricanes coach Peter Laviolette.

Trotz, generally considered a defensive mastermind, has the Capitals playing something that has the faint whiff of defensively responsible hockey to it.

Washington’s CA/60 has improved from 56.9 (23rd) to a middling 52.9 (15th). On the other side of the ledger, the Caps’ CF/60 has improved from 51.9 to 54.8.

Whether it’s Trotz or off-season personnel changes, the Caps have gone from horrendous 5-on-5 play to respectable mediocrity. And yet ... the team isn’t winning many more games.

Why? Adam Oates had better team goaltending (5-on-5 Sv% of .925 last season vs. .915 this season) and a few more power play opportunities per game. But if you rush to conclude Trotz is an unheralded genius, look at the Predators. Under Laviolette, Nashville has taken a titanic step forward, improving its GF% by 16.7%, its Point % by 15.3%, its CF/60 by an absurd 7.7, and its CA/60 by 1.6. The Predators are simultaneously more exciting to watch and better defensively than they were under Trotz.

Was Trotz holding them back? Or is this simply a combination of young defensemen like Ryan Ellis and Roman Josi continuing to develop, coupled with the emergence of Filip Forsberg, and a healthy Pekka Rinne?

Then there are the Pittsburgh Penguins, whose off-season roster changes seemed like change for its own sake rather than improvement (James Neal and Matt Niskanen shipped out; Patric Hornqvist and Christian Ehrhoff brought in). If you told Mike Johnston this summer to expect Sidney Crosby to be on pace for a career low in goals and sidelined with a case of the mumps in mid-December, I’m not sure he would have unpacked his boxes.

Nevertheless, by every measure the Penguins are a much better team and on pace to finish with a better record than one that was sixth-best in the league last season. Johnston, who has publicly embraced analytics, has done a terrific job so far.

Coaches’ results depend on their players’ ups and downs, so it’s not always easy to assess who should get a second chance. Nevertheless, teams have to make decisions about whether to retain or fire coaches somehow, and analytics can tell them a lot more than simply looking at the standings.

The Department of Hockey Analytics employs advanced statistical methods and innovative approaches to better understand the game of hockey. Its three founders are Ian Cooper, a lawyer, former player agent and Wharton Business School graduate; Dr. Phil Curry, a professor of economics at the University of Waterloo; and IJay Palansky, a litigator at the law firm of Armstrong Teasdale, former high-stakes professional poker player, and Harvard Law School graduate.

Copyright © 2014 by Ian Cooper, The Department of Hockey Analytics

Distributed by Torstar Syndication Service