Although advanced stats have been a point of discussion among hockey enthusiasts for several years now, it hasn't been until this summer that the enthusiasm materialized in mass hirings of statistical analysts and even whole statistical departments.
While the NHL teams (some more than others) have long dabbled in statistical analysis, the hiring of several advanced stats proponents serves as a clear indicator of the movement becoming accepted by the mainstream. The NHL has always trailed other sports in this regard, so chalk one up for the stats crowd. As The Boston Globe points out in its NHL On The Brink Intelligence Explosion piece: Edmonton hired Tyler Dellow, a former lawyer and stats-centric blogger, as a statistical consultant. New Jersey landed Sunny Mehta, an ex-professional poker player, as its director of hockey analytics. Eric Tulsky majored in chemistry and physics at Harvard and has a PhD in chemistry from Berkeley. Brian Macdonald was an assistant professor in mathematical sciences at West Point. Tulsky and Macdonald are consulting for undisclosed clubs.
That said, the one question worth answering is whether the NHL's rather icy attitude towards implementing statistical analysis stems from reasons other than an old-boys network refusing to change its habits.
One factor and perhaps one that doesn't receive enough attention is the inherent difficulty of performing statistical analysis in a sport as complex and organic as hockey. The game of hockey is random to a point and keeping possession of the puck is in large part responsible for winning. That's not exactly new knowledge and one hardly anyone involved with hockey at any level would dispute. Though, credit has to be given to advanced stats for providing actual numerical values for some of these possession related factors.
With the increasingly wide-spread acceptance of statistical analysis there has been a tangible shift in the way hockey is thought of and processed in the minds of the hockey audience. Like with all mentality shifts it also comes with a slew of odd overzealous beliefs of its own. There is an increasing mentality that champions an approach based on discarding anything not measurable as luck, random, or white noise. The statistical approach was built on people who were willing to think about the game of hockey in new ways. Yet, as the statistical approach gained acceptance and entrenched itself in the mind of the mainstream, it has also developed idiosyncrasies and mental rigidity just as strong (and at times ridiculous) as the traditional methods it often argued against.
On a micro-level there are a slew of things that a pure statistical approach to the game would peddle off as simply random or a small sample size (and therefore irrelevant). Maybe someone performs better under pressure, maybe someone reads the opponent better as the series progresses, maybe someone changes the tone of his game from a skill-based game to a more physical iteration, sensing that the opposition is responding poorly to continued physical play by being battered up. Maybe a goalie notices the opposition is targeting a particular part of the net due to a scouting report and adjusts accordingly, maybe it only means he had one great game from that alone. Maybe a player decides to play more conservatively to aid his injured defenseman with break-outs as he's unable to utilize his full mobility. What if someone that is considered "clutch" (a non-entity in statistical world) is simply better at adjusting to a 7 game series? Maybe he knows that the opposition's defense is getting battered up and he can score from a rebound in close so he attacks the net more than usual. In a game or a playoff series all of that would be detected as random by statistical analysis because the sample size is too small. There is a myriad of things that go into it that would simply be peddled off as luck by statistics and yet it’s become en vogue to pretend these things don’t exist.
From the very same thought process also stems the often found derisive attitude towards winning. It’s become somewhat prevalent to engage in hypothetical player-swaps where player A would surely win by playing on player B’s team, as long as he’s perceived as equally talented. Winning gets shoved aside as a magical occurrence where one is bestowed by luck to play on a great team (nevermind that he might be the team’s best player). That line of thinking is particularly dubious since the only point of professional sports is winning. Key players who have shown the ability to win while being number one at their position have basically shown that they’re able to achieve the only point of their profession. Two players might be equally talented but why isn’t winning a valid argument for one player over the other? We might not know what player A is capable of if he were playing on a better team, but we already know that player B is capable of winning in that slot and have no tangible evidence for player A being capable of the same feat. No matter how reasonable that assumption might be, it’s still an assumption and an analytical approach apparently favors hard evidence, doesn’t it?
To stop with the micro-analysis, where advanced stats are particularly limited is in the macro-look – the team building aspect. The advanced stats are not able to capture what players like Michal Handzus or Robyn Regehr can do in terms of setting an example and being part of the on-going effort to set an identity and the style of play a particular team wants. That’s why Calgary can afford to play hard-working players of limited talent. They aren’t winning anything as it is, why not set an example for the standard that is expected so that their youth who does actually possess talent sees what is expected in terms of conduct and professionalism? Are the 5 extra points in the standings and marginally more talented players worth more than having a group of hard working professionals who embody the standard that you would expect your talented players to subscribe to? Did the two-way responsible game with no cheating come out of nowhere to Anze Kopitar? Did the take care of your own end first and leave some points on the table come naturally for Drew Doughty? There was a standard and expectation that everyone followed, and it was nurtured by the careful selection of players that surely weren’t always the best most talented statistical fits.
When the best player on the team plays like a 4th line grinder in danger of being sent down the next day it sets a standard. How do you statistically quantify that aside from winning? Jonathan Toews has that in spades. Is he as talented as Alex Ovechkin? No. But that’s the difference between a great player that wins and merely an exceptional player. With that standard of play comes accountability and with accountability comes a cascading effect throughout the line-up. How can you afford to cheat in any aspect of the game when your best player plays at an elite level and gives maximum effort in all 3 zones? You can’t. That’s why Toews type of players will never be completely statistically measurable, their measure comes in winning and the standard they set for the team to follow.
There is no doubt that statistical analysis is a great tool to add to anyone’s arsenal for player evaluation. The mainstream acceptance of statistical tools is a big step forward for hockey, unfortunately as a side-effect it’s become almost a norm to be over-zealous in its use and derisive towards things that still matter and always will matter in hockey. The fanatical belief in analytical tools won’t lead you any closer to understanding what leads to a Stanley Cup winning team than a purely eye-based observation will. Both have its blind spots, and both are best used as one of the tools in your belt. Sometimes you need a hammer, and sometimes you need a wrench, it's best to carry as many tools with you as possible and utilize them accordingly.
Post-Series Talk: Edmonton Oilers defeat the LA Kings, 4-1
