Between the ongoing Conference Finals and the media circus surrounding Mike Babcock’s hiring by the Maple Leafs, it’s been easy to overlook some solid analytics work being published these days. Below are some links and observations on recent analytics stuff in the NHL:
- Late last month, David Johnson posted this piece over at Hockey Analysis, in which he asked whether we were starting to see the on-ice results of the so-called Summer of Analytics, and Gabe Desjardins raised a similar question on Twitter this week. Both concluded that, contrary to some observers’ expectations, there hasn’t been a noticeable effect of last summer’s analytics hires. I would agree, though I’d also point out that the results we’ve seen have been consistent with my expectations. As a more general thought, analytics as they stand today have far more negative utility than positive utility; that is, it’s harder for us to tell teams how to get better than it is to tell them how to avoid getting worse. This isn’t intended as a slight against analysts; a team could hire an analyst to do nothing but veto bad trades and signings, and he/she would likely save that team many multiples of his/her salary. But the damage had already been done to teams like Edmonton, New Jersey, Carolina, and Toronto by the time they hired prominent analysts, and the road back to contention was always going to be difficult for those teams.
- Also prompting some discussion of the role of analytics in hockey was the hiring of Ray Shero to replace Lou Lamoriello as GM of the Devils. The Jersey blog In Lou We Trust had this take on Shero’s work in Pittsburgh, while the Driving Play blog offered an interesting counterpoint to some of the criticism of the move. While few would characterize Shero as an true believer in hockey analytics, I think that, for the most part, this was an excellent hire. In his time in Pittsburgh, Shero’s good moves (acquiring Marian Hossa and years of Pascal Dupuis for spare parts in 2008, acquiring Chris Kunitz for Ryan Whitney before the latter’s game fell off a cliff, signing Paul Martin to an excellent free-agent contract, swapping third-pairing puck-movers in Alex Goligoski and Matt Niskanen while adding sniper James Neal, acquiring Jarome Iginla at the 2013 deadline for fringe prospects, signing Tomas Vokoun to a backup salary, getting a year-plus of Jussi Jokinen with retained salary for a low draft pick) far outnumbered the bad. As far as the bad, I have a hard time dinging Shero too heavily for moves that 95% of other NHL GMs would copy, such as Marc-Andre Fleury’s seven-year deal. For me, the only truly indefensible move of his tenure in Pittsburgh was the decision to bring Rob Scuderi back, and even that may have been influenced by Mario Lemieux’s famous loyalty to former Penguins.
- When the Calgary Flames were eliminated in the second round of the playoffs by the Anaheim Ducks, it didn’t take long before someone (in this case, Garret Hohl) wondered whether they’ll follow in the footsteps of teams like the 2014-15 Avalanche, 2013-14 Maple Leafs, or 2012-13 Predators in crashing back to earth after unsustainably strong results this season. Given how poor this year’s Flames were at defending and driving play, it’s easy to imagine them experiencing some degree of regression to the mean next season. Still, it’s important to define what constitutes “crashing back to earth”. In the case of the Avalanche, they dropped to 21st in the league this season after having the NHL’s third-best record in 2013-14, a fall of 22 points in the standings. A similar drop by the Flames would leave them with just 75 points, as they finished 16th in the NHL this season and barely made the playoffs. In other words, they’d have to end up as one of the worst teams in the league in 2015-16 to have collapsed as badly as the Avs did this year; falling to, say, the wrong side of the playoff bubble wouldn’t actually be much of a crash.
- Speaking of work at Hockey Graphs, this piece on shot location data from Ben Wendorf is very worth reading.
- More interesting game-tracking work continues to come out every week. Jennifer Lute Costella recently published a very detailed breakdown of defensive zone exit data from the Blackhawks-Predators series, while Ryan Stimson released the first post of what should be a long series on passing data.
- It’s starting to feel a bit like piling on, but Michael Lopez has an excellent critique of the NHL’s “85% accurate” playoff predictions model.
- From earlier this month, this piece on momentum in NHL hockey is great.