When I left sports media to join an American Hockey League coaching staff in November 2017, I felt like I had a pretty good grip on how to evaluate players and teams. I grew up around the NHL and eventually played up to the AHL myself, and I was always a player who wanted to understand our systems thoroughly before I went over the boards, so I had a good grip on what’s supposed to happen out there.
Having spent years after my playing days watching and talking about the league for a variety of outlets, I had come to trust the ol' eye test too. To go with that, I had enough fear of the analytics movement to recognize I should either force myself to learn what the numbers meant or risk getting left behind. It was that mix that got me the job in the first place, and I was eager to put my abilities to the test in the real world.
It wasn’t long after I started – a month into the season or so - that I saw our first coach’s packet from the analytics department, which presented us with everything we needed to know about our upcoming opponent and our own team over the season’s small sample to date. It was a little overwhelming at first, as pages of numbers can be when you don’t know what you’re looking at.
I remember seeing a number that showed that over the past few games, our zone exit percentage was ... well, whatever it was. If I told you it was 62 per cent, or it was 28 per cent or really anything outside the most extreme percentiles, most hockey fans probably wouldn’t blink. Because, y’know ... what’s good anyway? It was one of those moments where I was stuck in my own head, new in the coaches' office, wondering: “Should I ask if that’s good or bad? Would that look dumb?” I finally had the type of internal information in front of me I had been excited to see but didn’t know what it was telling me.
Of course, the only dumb thing would be not asking, but I wasn’t pumped about exposing myself so early. I ended up trying to make a joke about it to get the conversation going, and in the end, asking wasn’t dumb at all, because others had questions too.
Eventually “league average” and “team average” were added to a variety of numbers for quick reference points. With statistics, context is everything. Who had we dressed in those few games (did we have injuries?), who had we played (were they a good forechecking team?), what’s typical, in many cases, the answer to “Is that good” doesn’t even exist as a yes or no without more information.
It was a small example of a major challenge in data analysis that’s gotten much more difficult around the NHL today: Here are the numbers, but what do they mean, what useful bits of on-ice information can be pulled from these oceans of data? I say “data analysis today,” but I’m sure that’s forever been at the root of statistical analysis, it’s just that the NHL is at a crossroads of information where more advanced numbers are commonly accepted across 32 teams (to varying degrees), but there’s just so damn much of it coming into existence right now, so do you have the right people to find the relevant context to read the proper meaning into those numbers?
It’s the “so damn much of it coming into existence right now” part that I’m focusing on here today. I left the Marlies to spend more time around my growing family before the 2018 season, and since then, there’s been a boom in available information, thanks in large part to the emergence of quality tracking data.
NHL teams have some amount of proprietary information and have built their own internal databases to pick through. Most teams have had that for years. But more recently the NHL has given them access to the player tracking numbers they’ve been working on, with far more coming as puck tracking hopefully integrates smoothly into gameplay in the year ahead. At first, that mountain of information from the league would’ve been daunting and so the question will be asked of teams: do you have people who can make sense of it? Do you have people who will misinterpret it or not even bother with it? Can you learn anything good from this information aside from neat trivia like, “Did you know Connor McDavid skates fast?”
It reminds me a bit of one of the first windows for improvement teams had when analytics first emerged as relevant for teams from the Great Corsi Wars of almost a decade ago. The early adopters got an advantage, the next wave didn’t generally fall behind, and the last accepters, well, they had fallen behind. Similarly, right now it’ll be about the quality of people teams have internally, not whether they have people or not. There’s useful information in tracking data, and those with good people will get ahead, those in the middle won’t fall behind, and you know what comes after that.
Further to the NHL’s tracking data, there are companies like SportLogiq who do their own tracking and many NHL teams have signed up to get their hands on their information. While the company provides them access to their database, teams are free to do what they like with the numbers once they’ve paid for them.
Some may integrate it into their internal systems, some may use it as it's provided, but ultimately what it comes down to is, “What answers can you find within and can those answers help your team get even a small percentage better?”
To go along with this data is the improved access to global video, meaning there are few players around the world you can’t watch play. Teams now cannot just rely on a few scouts or pure stats to assess the progress of international skaters. That’s only a good thing if the additional eyeballs you have on those games are any good at scouting. You’re seeing a theme here.
More than ever, NHL teams will be punished for hiring buddies, while those who seek out legitimate front office talent should thrive in the decade ahead. It’s impossible to perfectly predict how people and teams will develop, but in theory anyway, there should be less abject guessing in the years ahead.
There are no more “analytics” teams in the NHL, as all organizations embrace increased volumes of information. What there will be, are teams who excel at handling the oceans of it they’re being given, while others do their best to simply tread water, and end up going under.