March 14 2017 10:00AM
This past Saturday, I descended into the bowels of Rogers Arena through Gate 9, a small entrance below the Georgia Viaduct reserved for staff and security. In the elevator, all of the buttons were embossed with a green and blue killer whale. As I made my way to the fourth floor of the building, to a room aptly named The Sports Bar, I was reminded ad nauseam that #WeAreAllCanucks.
I was at the Vancouver Hockey Analytics Conference and this is what I learned.
1) Enough of "Us vs. Them"
Sportsnet's Dimitri Filipovic and Dan Murphy began the day with a panel moderated by Garret Hohl on analytics in broadcasting.
"I think there's a future NHL GM in this room." - Jon Wall
Murphy began the panel with an anecdote about when Rogers first took over the national NHL broadcasting rights. If you remember, Darren Pang used to stand in front of a large touch screen where he used puck-shaped devices to reveal various statistics in between periods and games. Apparently, the overall public reaction to this was so negative that the entire segment was scrapped.
"are millennials killing traditional hockey TV analysis???" #mycolumn— Josh Weissbock (@joshweissbock) March 11, 2017
He advocated for an introduction of concepts before numbers in MSM broadcasts. If broadcasters can first begin to describe the ideas behind various statistics in layman's terms—for example, WOWY as described as one player making everyone else crap when they're on the ice with him etc.—then the numbers themselves can be introduced later. Basically, let's get everyone on the same page before scaring people away with math.
Murphy argued that, ultimately, the "us versus them" attitude among analytics types and the mainstream media needs to end for the media to embrace and promote advanced statistics on the ever-important mainstream platforms.
2) The Kelowna Rockets Analytics Department Has a Sense of Humour
Mark Francis of the Kelowna Rockets organization talked about the importance of appealing to younger players through humour when trying to implement the work his department does into the on-ice product.
For instance, the Rockets analytics team was measuring effort plays, specifically shots and hits, early in games and wanted to inspire their team to register more of these events early on. What they came up with was a metric called SHITS (shots + hits).
The result was a locker room full of adolescent men motivated to give as many SHITS as possible in the opening minutes of the hockey game.
3) So Does Former Rocket Tyson Baillie
Francis also talked about a particular incident when his group was trying to implement a tactical change during an intermission. They relayed the new offensive zone strategy to the coaching staff and what resulted was a goal from Tyson Baillie.
Skating back to the bench, Baillie yelled to his team, "YEAH ANALYTICS!"
4) What Does It Mean to Draft Perfectly?
This was the title of Namita Nandakumar's talk in which she sought to describe what it would mean for a team to draft perfectly in any given year.
She assigned former draft picks a career value and then used this to determine the Draft Efficiency of teams during the 2000's, where Draft Efficiency=Actual Draft Value (sum of players picked by individual team)/Perfect Draft Value (sum of their best possible draft in that year).
5) Measuring Grit
Social Scientist Stefan Wolesjszo presented on how to apply his field's work to an NHL roster in an attempt to measure the many "intangibles" we talk about when we talk about puck.
He used five criteria to evaluate the value of each of these measurements when translated to hockey: Conceptual Clarity, Empirical Research, Elite Occupations, Practicality and Applicability.
The five intangibles he looked at were: Cognitive Fatigue, Resilience, Leadership, Motivation and Grit.
In the end, he believes Cognitive Fatigue and Resiliency would be best suited for evaluation in an NHL setting, and that Grit and Resiliency are more or less the same.
If and when these measurements are implemented, the question for NHL teams when building a roster will be: How to measure the trade off between Skill and Grit?
6) Flurry-Adjusted Expected Goals
In his metric, Tanner tries to adjust for the likelihood of scoring in a flurry of shots—that is, the effect of the time between shooting events on xGF.
To avoid butchering an explanation of his work any further, you can get the full low down from the man himself at MoneyPuck.com.
7) Brian Elliott Double Dips in Salsa
...and is therefore a bad person.
Get at Chris Watkins (@yolo_pinyato) for a full explanation.
8) Pressure and Score Effects
"Treating continuous variables as if they are discrete makes garbage." - Micah Blake McCurdy
Stats scene favourite Micah Blake McCurdy spoke about considering pressure as an independent variable in lieu of time to get at what is responsible for Score Effects. He broke this down into defensive pressure and offensive pressure: how much you have to lose when you're scored on, and how much you have to gain if you score. These can be understood as "Expected Standings Points Lost" and "Expected Standings Points Gained".
Micah found that the home team shoots more than the away team when down by one or tied, and also score more in these situations. In a tie, defensive pressure increases (in an effort to force overtime and gain at least one standings point) while offensive pressure decreases.
In general, the trailing team produces more shots and goals than the leading team, and the home team produces more than the away team in these situations.
9) The Dump and Change
Joshua Smolow followed up on the work being done on zone entries and focussed on one of the more familiar strategies in hockey, the dump and change, while isolating it as a zone entry tactic not meant to create offence (as opposed to dump and chase or a carry-in, for instance).
While the sample of games he's looked at so far is still small, he found that there were an average of 13.8 dump and changes per game and that only 10% of them resulted in that team maintaining possession in the offensive zone. While >60% of DnCs resulted in offensive zone entries made by the other team, a DnC was actually less likely than other zone entries to result in a controlled zone entry made by the opponent (when the opponent is next to execute a zone entry).
He hypothesizes that this might be as a result of the team that dumps and changes being able to regroup in the neutral zone and prevent a controlled entry.
10) Talent Distributions Within NHL Rosters
Alex Novet gave an excellent talk exploring hockey in terms of Strong and Weak Link Games. A Strong Link Game is one in which the team with the best player is most likely to be successful, and a Weak Link Game is one in which the team with the worst player is most likely to lose.
Hockey has qualities that apply to both. While it's a game where there are few players on the ice at once and the time on ice of weak players can be limited, it's also a game where one goal can decide a game and the best players still have limited time on ice.
Ultimately, Novet's findings suggested that hockey is more of a Strong Link Game, resulting in the following implications:
a) When it comes to trading, choose quality over quantity.
b) When it comes to contracts, spend money on the star, not on the bench.
c) When it comes to drafting, tanking works.
d) When it comes to tactics, hockey is more about creating plays than avoiding mistakes.
Alex writes for Hockey Graphs and has made a public promise to write a post on his incredible #VanHAC talk soon. Stay tuned.
11) Meghan Chayka is Awesome
The #VanHAC keynote speaker this year was Meghan Chayka, the co-founder of Stathletes, an independent resource for hockey analytics. If this business, statistics and tech guru's name sounds familiar, it might be because her brother, John Chayka, is the GM of the Phoenix Coyotes.
That said, Meghan is far from living in her brother's shadow and has made her mark on every industry and field she's entered. She took the opportunity as keynote to delve into the representation of women in tech, statistics and business.
She stressed that opportunity should not discriminate—or in her words, that the ball should bounce the same for everyone. She also spoke about how data should not only be used internally, but also to create new fans. That's to say that we should use analytics to empower hockey fans to understand the game better.
Meghan (@MeghanChayka) left attendees with a list of influential women in her various fields to follow on Twitter, so I thought I would share them with you here. I encourage you to follow each and every one of them.
12) We're Still Not Over the Taylor Hall Trade
Somehow we made it to the penultimate talk before the swipes at the hall trade came out.— Micah Blake McCurdy (@IneffectiveMath) March 11, 2017
A huge thank you to all the presenters and organizers for a most awesome day. And also to whoever made these sandwiches.