For the first time in the history of professional hockey, the nerds are winning.
With at least four NHL teams hiring advanced stats gurus in the past six months, it seems that the #fancystats movement is just about to reach critical mass both in NHL front offices and in the mainstream media. Really, these are exciting times for bloggers or anyone interested in seeing sports through a stats-oriented lens.
And it’s exactly why now is the best time for the game’s brain trust to dump everything and start from scratch.
NHL hockey is a zero-sum game: there’s a loser and a winner in every single one of the league’s 1230 regular season game every year. That number is not set to rise significantly for the foreseeable future.
Other fields like finance, technology and education are non-zero-sum games. Innovations in those fields are like rising tides which lift all ships. You’re assured of a payoff if you embrace change, even if your relative standing in your field does not improve.
All this is to say that NHL teams hoping to gain an edge by embracing known metrics such as Corsi or player usage charts are not assured to actually do better on the ice. Just "keeping up" in the NHL brings you zero additional benefit, and even becoming legitimately better as a team through sound moves will only pay off about 60% of the time (the rest of the time, you lose because you’re unlucky, or the San Jose Sharks…). The competitive advantage of using RTSS-sourced databases (Extra Skater, Behind the Net, stats.hockeyanalysis.com) is rapidly evaporating. Yesterday’s undervalued players like Matt Niskanen, Anton Stralman and Benoit Pouliot are tomorrow’s overpaid supporting actors. Information that everyone has tends to become worthless in a hurry.
Fast-developing technological powers such as China and South Korea hop-scotched First World Countries in terms of economic growth by bypassing the desktop age entirely and embracing smart phones a decade before North Americans did. Wise NHL execs may be well served by taking a page out of that playbook and looking beyond the most-hyped metrics today entirely, in an effort to find the next breakthrough, and the next, and the next.
Or as Wayne Gretzky said, "I skate to where the puck is going to be, not where it has been."
Rather than going too much into details into where I think the next analytics frontier lays, I’ll wrap up with a few questions to consider for people looking to build a better hockey team:
What are some reliable genetic indicators suggesting future NHL success?
Is it possible to "walk into" an NHL starting goaltender by signing the best 3 undrafted goalies each year from the NCAA, keeping the one posting the highest save percentage, and repeating the process annually?
Are offer sheets a under-utilized tool?
What technological tools can be used to fully exploit the quantum leap in data quantity associated with the implementation of SportVU?
Which SportVU-tracked metrics are repeatable and correlated with future success?
How can video scouting be made a less time-consuming task? Is there a scalable solution in turning raw game footage into sortable data?
What types of player should you OVERPAY at the NHL level?
Should all players aim to play 82 games in a regular season?
How can wearable technology such as Nike Fuel and Google Glass change training techniques and in-game coaching?
How can NHL front offices attract and retain the best management and technical talent, knowing that ONE Costco store creates more revenue per year than an NHL franchise?