Recently, I’ve received some unsolicited emails from some very smart young people about working in analytics for a hockey team. There are definitely people more qualified that they could’ve tracked down, but most of them are not allowed to talk about their jobs, so I guess they were stuck with me.
It felt a little bit strange corresponding with these mathematics or engineering students, because theoretically, their backgrounds are a lot more suited to this line of work than mine (I graduated in Marketing). I apparently passed Calculus II 10 years ago (I barely remember taking it). I’m a mediocre programmer. And I don’t even work in the NHL.
But there are still a few thoughts I could share.
In search of lost time
I’m now in my second season as Video & Analytics Coordinator for the McGill Martlets, working for coach Peter Smith (who has won CIS National Championships, as well as gold with Team Canada at the Olympics), and I’ve come to realize the number one demand of the job isn’t math skills, but time management skills.
Considering that there are only 24 hours in a day, it’s always better to run a fast, focused 45-minute practice instead of spreading out those reps in a slower-paced 90-minute session. That’s how our coaching staff does it, and it gives our players a little extra time each day to study, socialize or catch up on sleep. Over the course of a year, it’s a significant advantage.
Running these short, intense practice is not my responsibility – our coaches have been doing thing this way forever. But, over the past year, I’ve started to apply that line of thinking into the way we address video and statistics.
The 100-10-1 Rule
Peter puts tremendous value on preparation. On the one hand, he loves taking in information – he spends a significant chunk of his day looking at game tape or reviewing stats. On the other hand, he is very careful about the quantity of information that he passes along to the players, for fear of overwhelming them or diluting his key message with other things that don’t matter as much.
So during the time we’ve worked together, I’ve developed an informal time management rule: 100-10-1
What it essentially means is that for every task he delegates to me (whether is it is coding a game’s footage, compiling our advanced stats for the last five games, or anything else), I aim to invest 100 minutes of my time, so that it takes Peter no more than 10 minutes to prepare a coaching message which he will deliver in one minute (or less) to the team. The key here is for me to do a lot of work in an organized manner, and then help Peter draw out the essentials insights.
Doing more with less
The 100-10-1 rule has one serious drawback: it shifts a lot of the burden onto me (the person in charge of doing the legwork). We therefore needed to find tools and techniques to make the 100 a smaller number, or else I’d be quickly overwhelmed. (In addition to my responsibilities at McGill, I also work a 40-hour per week job)
To survive the season with my sanity intact, I need to make my work scalable (meaning that if I put in 10 minutes on a specific task this week, I won’t need to put in 10 minutes again the week after). I needed better software solutions than what most other assistants or video coaches have been relying on.
For instance, in an NHL game, two league employees sit in the press box and mark in-game events (RTSS), two team employees record and code the game for the coaching staff, and a team analyst (or some random blogger at home) crunches the numbers after the game. That’s about 15 man-hours per game, which is way too much for my purposes.
So, at the start of last season, I reprogrammed our video capture and analysis software so that I can track advanced stats in real time, then dump it into a spreadsheet that I can sort out after the game. Now, a single person (myself) can do all that work in about three hours, which opens up a world of possibilities in other aspects.
Because of the demands of my job, I don’t actually do a great deal of pure research. Roughly 60% of my time at the rink is spent watching, coding and organizing game tape, which is something a traditional video coach would do, 20% of my time is spent on gathering stats and “doing” analytics, 10% of my time is spent operating various IT productivity tools, and the remaining 10% of the time, I’m on the ice helping with practice or shooting on our goaltenders.
I suspect there are a lot of very deep thinkers out there who would be disappointed to work for a team. It’s one thing to spend your spare time scraping play-by-play data and investigating things that interest you, but it’s quite another thing to develop a “customer-oriented” mindset and doing things which may not necessarily make sense to you at first for people who don’t know much about your approach. Peter is my boss, but mostly he’s my client – you gotta make your client happy.
As an example, one of the first things that Peter asked me to track last year was the number of turnovers his players were making. I was initially resistant to devote time to that, because, as we know, turnovers aren’t always a bad thing – the more you have the puck, the more you’re bound to turn it over. But we ended up having a productive back-and-forth on the matter, and found a slightly different way to account for turnovers (both statistically and semantically), which turned out to be very useful for us. Instead of standing my ground (“turnovers are a stupid stat”), I tried to serve Peter, and we created some proprietary knowledge because of that dialogue.
Coaches knows a lot about hockey and “stats guys” (like myself) think they know a lot, so it’s not always easy to sit down and hash things out. But an ability to take the time to find common ground with those who disagree with you is a skill any aspiring analyst should attempt to acquire.
(Note: I will be talking about IT & Knowledge Management For Elite Hockey Organizations at the RIT Hockey Analytics Conference on October 10th, 2015. Click here for more information)
Jack Han is the Video & Analytics Coordinator for the McGill Martlet Hockey team. He also writes occasionally about the NHL for Habs Eyes on the Prize. You can find him on Twitter or on the ice at McConnell Arena.
2 thoughts on “It’s not about the numbers: working in analytics for an elite hockey program”
I love this, cause it reflects exactly the juggling I did when I was in that role for a few teams.
sincerely, a former McGill Redmen stats guy. 🙂
As they say, “the more things change…”