The NHL’s Numbers Game: The Evolution of Hockey’s Analytics Movement

The subtitle of Michael Lewis’ legendary Moneyball is “The Art of Winning an Unfair Game.” The unfairness, of course, being the disparity between obscenely wealthy clubs like Yankees and Red Sox and slightly less wealthy ones like the Oakland Athletics.

If you wrote a book about the burgeoning hockey analytics movement, you might give it the same subtitle, but for different reasons. Not only is hockey a cruel, often brutal game, but at times it doesn’t make any sense, at least from a basic statistical standpoint: A team could get more shots on goal than another, build an almost comical advantage, and still lose. At heart, hockey is still 12 brave souls skating at top speeds inside a closed venue, so the game is often just random and, well, unfair.

Analytics, advanced stats, fancy stats — whatever you want to call them — have slowly-but-surely started to fight their way into hockey, through means both sensational (“traditional” TV analysts callously smacking down those that differ from their thinking) and subliminal: almost every NHL franchise uses some form of analytics, and more and more teams are bringing in stat-practitioners to staff entire analytics departments. Hockey’s most recent offseason was half-seriously referred to as the “summer of analytics,” after teams like New Jersey and Toronto made waves by hiring everyone from up-and-coming junior execs to poker pros.

The Devils and Maple Leafs are important, because they’re two teams currently at opposite ends of the advanced-stats spectrum. Last season, New Jersey was one of the top teams in CF percentage (“Corsi For,” which counts total shots – whether they reach the net or not) and Fenwick (which tallies shot attempts minus blocked shots). The former is a good indicator of puck possession – the higher the number, the better – and the latter measures stress on a goaltender, and given that the Devils were near the top of the league in both, most were mystified that they missed the playoffs once again.

On the surface, you could blame it on their decision to start legendary, but past-his-prime goaltender Martin Brodeur over the now-superior Cory Schneider. But a deeper look suggests it went beyond that.

“The biggest problem for the Devils over the past few seasons has been their offensive system,” Jen Lute Costella, an attorney-turned-analytics-writer for Yahoo Sports’ Puck Daddy, says. “Not only did the Devils’ systems lead to their opponent’s shots being suppressed, but their own as well.”

Even if their CF percentage or Fenwick was high, they weren’t getting enough shot attempts, forcing them to play on the razor’s edge: each game could be lost by a bounce of the puck or sporadically bad goaltending, which Brodeur (who despite his reputation, remains unsigned and a free agent at 42) was often more than capable of supplying.

So what did the Devils do? They gambled – literally. In June, New Jersey hired poker pro Sunny Mehta to head up their analytics department, a move that highlighted what a lot of experts want to see from this movement.

“To me, the whole point of the analytics movement in hockey is to play a higher quality of hockey,” Costella says. “As teams get more advanced and savvy in the way they analyze the performance of the team, they can use their players in ways that they may not have thought to do in the past.”

And Mehta has already begun to address the Devils’ deficiencies. They’re currently second in their division, and not only are they consistently starting Schneider, “they have increased the rate at which they are getting shots off a little bit,” Costella says, “which they will have to continue if the team has plans on seeing the postseason again.”

While New Jersey has been quietly going about their business, Toronto has become a battleground for hockey’s “old school vs. new school” mentality.

The Leafs, despite a playoff appearance in 2013, are a team that “ranks in the bottom five in puck possession, so this is still a team that will get outplayed, outshot, and outchanced most nights,” according to Rob Vollman, author of the essential Hockey Abstract. The previous solution was to keep the team — led by a coach in Randy Carlyle who approves of “tough guys” and “hitting” — loaded with bruising fourth-liners like Colton Orr.

One of the bugaboos of the fancy stats crowd also happens to be one of hockey’s most entertaining acts: the body check. While yes, it is exciting, the fact is, players who get a lot of hits don’t typically have the puck very often, since their main purpose is more or less to punish their opponent. Yet Toronto remained a proudly physical team – and defiantly anti-analytics.

“Putting out linebackers on skates may make you ‘truculent’ and ‘tough to play against,'” Timo Seppa, owner and editor of the annual Hockey Prospectus, says. “But it won’t do much to help you win games.”

“Toronto’s whole anti-analytics stance was laughable, but it was only a portion of the problem with the team over the last few seasons,” Travis Yost, who Canada’s TSN hired as part of their new analytics department, adds. “Reality is, the team wasn’t just getting submarined by an old-timey head coach and a front office that rebuked any consideration of statistical analysis. In years past, there was a glaring depth issue in the forward ranks, and defensively, their top-pairing was getting caved-in far too often.”

The response – led by the new regime of president and Hockey Hall of Famer Brendan Shanahan – was to finally give the smart guys a seat at the table, with the headlines going to Kyle Dubas, a 28-year-old GM of a Canadian junior team who is a massive proponent of metrics. Seppa described him as “Theo Epstein-esque,” and correctly pointed out that this is a strength Toronto, one of the NHL’s richest teams, should be able to exploit.

“Toronto may have the same cap space as every other team,” he says, “but they should be able to trounce most of the league in analytics investments, should they chose to – and they seem to be. Credit Shanahan for getting the ball rolling there.”

That summer, the team signed players like Daniel Winnik, Leo Komarov and Stephane Robidas, not household names, but “together they show the team’s willingness to consider alternatives to the team’s usual design,” Vollman says. 

“They’ve made a considerable jump from woeful teams of the past,” Yost adds, “but you can make a big improvement and still be a bad team. That’s sort of the issue with Toronto right now. They still have a number of issues to figure out – fixing center depth and getting a player like Jake Gardiner or Morgan Rielly to really develop into a reliable first-pairing guy would be huge in terms of making another step forward.”

Toronto’s an important market to single out because it’s where most of hockey’s media comes from. Similar to baseball before its stats revolution, hockey’s front offices and television analysts make up an old boys’ network that doesn’t like being challenged, and the summer of analytics has led to some scoffing from major NHL TV figures. NBC’s lead hockey analyst, Pierre McGuire, said that any coach who relied on analytics to evaluate players “should be fired.”

However, the only people that should really be interested are front office types. Seppa, warns that “Among those folks working in the game, it would be foolish to not at least get a working knowledge of some of the key concepts. There are many holdouts, even among GMs. For an owner, that should be troubling.”

Analytics are here to stay, however, and not merely a fad. They simply make too much business sense. Much like Moneyball, they can help shine a light on players who don’t fit into the typical narratives, improve team play and – most important of all – impact the win/loss column.

“Even if hockey analytics give you a 2-5 percent advantage in terms of acquiring talent through the draft/free agency/trade, identifying superior strategy and systems play, optimizing lineup deployment – it’s still a competitive advantage,” Yost says.

“For teams, it provides that edge, much like they have received from their trainers and equipment managers,” Vollman adds. “It’s that sober second thought, and the ability to find things that the traditional analysis has missed, that provides the greatest advantage.”

Plus, even if analytics are a fad, it’s becoming increasingly difficult to ignore the impact they’ve had on the game:

“The league’s two best teams, Chicago and Los Angeles, have been cashing in on that edge,” Vollman continues. “And unless they’re foolish enough to publish their secrets in a book like the Oakland A’s did, the rest of the NHL will need to build teams of their own to catch up.”