The coach in question has several international titles and an Olympic gold medal to his name, so it was an exceedingly interesting opportunity to experience a different perspective on the use of analytics and uncover a few reasons why old school hockey minds may be resistant to adopt them.
"No we don’t; yes I am."
Above was the email reply I received after asking whether he is currently tracking possession stats for his team, and whether he would be interested in learning more about advanced stats. It was a concise, if telling answer.
From old school to new school
One of the first things I noticed about the coach’s office, aside from the assorted hockey memorabilia hanging on the walls, was the fact that he had two laptops set up side-by-side on his L-shaped desk. One of the laptops was connected to a 50-inch flat-screen TV hanging from the far wall; on the screen was the game footage from his players’ last outing against an American collegiate team.
After getting acquainted, he walked me through the video analysis tool he uses to break down game footage, a specialized program that is very similar to the one used by the Montreal Canadiens and other NHL franchises. Users can create video snippets of important plays, whether it is goals, scoring chances, or saves, and send them to a database to create a playlist of teaching points. My host maintained that video is the best educational tool at his disposal when it comes to helping his players improve their understanding of the game.
Crunching numbers (or not)
Once we went over his video analysis process, the coach handed me a thick stack of paper which included shot location charts, faceoff win-loss data and pages and pages of hand-collected turnover statistics. For someone who claims not to use analytics, he and his coaching staff collect a lot of information. It would seem that at the highest level of the game, this type of attention to detail is the norm rather than the exception.
However, he is totally right when saying that he does not use analytics. It is one thing to meticulously track where turnovers happen on the ice and which player is responsible for them using pen and paper, but an entirely other thing to build databases and predictive models in order to find metrics which are persistent across seasons and strongly correlated to winning.
Very few hockey coaches have a strong technological, scientific or mathematical background, so it is natural that some of them would have trouble going from simply counting things to get a feel for what is happening on the ice to using technological tools (Excel, R, or another statistical package) to systematically track data over time in the hopes of uncovering some previously unknown trends.
A language barrier
All things considered, it seems that hiring someone to massage the wisdom out of the numbers may be the best move for most hockey organizations. However, the problem of communication between coaches and analysts remain. Since hockey is, above all, a people business, it is important for both jocks and nerds to be flexible and understanding of each other’s experiences.
For example, when it came to breaking down the powerplay, I mentioned that the most popular metric, at least in among bloggers, is Shots Per 60. Meanwhile, my host, who had never heard of the metric before, extolled the importance of shooting from prime locations and having good puck retrieval after each shot. Philosophically, we were actually talking about the same thing. but it took a few minutes for us to realize that it was only semantics which were separating us. Same went for a player’s individual Corsi versus his turnover numbers – it is really hard to maintain a positive shot differential when you are giving the puck away in your zone.
Contrary to what is portrayed in mainstream media, most "old school" professional coaches I've spoken to are tremendous students of the game who appreciate the importance of asking questions and collecting data to answer them. So why aren't all of those great coaches completely sold on the analytics movement? Here are a few hypotheses
1) They trust video more than spreadsheets - it is a better teaching tool, and "what you see is what you get."
2) They distrust complex formulas - the numbers and terms mean nothing to them, so they are not sold on their effectiveness.
3) They are too busy with planning practices (sometimes at 3AM), scouting opposing teams and drawing up plays to learn how to do multiple regressions - there are just 24 hours in a day, and focusing attention on a new task usually means neglecting another.
235) They *don't* think having better information is going to help them win more hockey games.