Sunday, August 10, 2014

Is this the apex of "analytics movement?"

Reprinted with permission from hockeybuzz.com


"There are three kinds of lies. Lies, damned lies and statistics."


American satirist Mark Twain used that phrase, possibly borrowed from British politician and writer Benjamin Disraeli who was said to have coined it.

A few years down the road from now as we sit at the possible apex of the analytics movement, will National Hockey League satirists and writers end up changing "statistics" in that quote to "analytics?"

The advanced stats trend is rolling forward full force to the point where the Edmonton Oilers earlier this week just hired a blogger to help out.

The Oilers have been in a slump since losing the Carolina Hurricanes in the 2006 Stanley Cup Finals. (to this point there's been no mention of a curse on the Edmonton franchise for being the first Canadian franchise to lose the Cup to a "non-traditional market" team like Carolina)


Tyler Dellow's website mc79hockey.com went dark on Monday night as a prelude to his newfound position as analytic guru for the Edmonton Oilers.

This follows an announcement back in July that the Toronto Maple Leafs bounced two long-time hockey guys Dave Poulin and Claude Loiselle in favor of an analytics guy.

New Leafs President Brendan Shanahan replaced those two assistant general managers with 28 yr. old Kyle Dubas.

Dubas is a young analytics guru whom Shanahan described as someone with "maturation beyond his years" bringing "a fresh approach will benefit our club for years to come."

As to where Dubas sees the Maple Leafs at this point? "It's not any secret if you watch our team you'll begin to see it clearly we don't really dump the puck in," he said. "We want our players to solve problems and make plays with the puck."

Apparently that was something that eluded Poulin and Loiselle, both of whom have been involved in hockey longer than Dubas has been alive.

After making the playoffs once since the 2004 lockout, and with a Cup drought dating back to 1967, something had to be done at "The Center of the Hockey Universe."

And it's not just our northern friends either who are jumping into this trend.

On August 1st, the New Jersey Devils started their own analytics department by hiring Sunny Metha.

Metha wrote two books on analytics within his given profession as a poker player.

That the Devils hired a professional poker player for their analytics department is apropos. Not only is New Jersey home to "Las Vegas East"--Atlantic City--but life without goalie Martin Brodeur looks be a crapshoot.

As for the Leafs Dubas, he seemingly doesn't consider himself a "stats guy," per se. He prefers to look at it this way, "I’m someone whose worked in hockey my entire life that over the past number of years has begun to develop a usefulness for analytics.  It’s been a big help to us (in Saulte St. Marie) in reducing uncertainty and just gathering more information.

How does Dubas do it? "You can kind of reach and track a bunch of different things that aren’t really mainstream yet and I think it all has value,” Dubas said. “It’s just a matter of … trying to find out which pieces of data can help us to eliminate uncertainty and which ones are maybe not what we thought they were.”

It all goes back to Moneyball, and the incredible work Oakland Athletics GM Billy Beane did with his baseball team.

By tracking statistics and analyzing them, Beane was able to draft and develop high quality baseball players who would help the team compete against the big money baseball machines like the NY Yankees and Boston Red Sox.

Those players, of course, would end up playing well enough to leave for the big paychecks in NY and Boston. But Beane was able to keep the team stocked and competitive for years by analyzing miles of statistics on prospects.

Pure analytics can work well in baseball simply because it's the most individualized of the four North American sports.

Just like a chess match where a computer can analyze and calculate moves and potential moves by one individual vs. one other individual, so can a computer measure one vs. the other in baseball as the game is essentially pitcher vs. hitter.

There's nothing remotely close to that in hockey (save for the penalty shot/shootout.) Unlike baseball where a hitter's average vs. lefties, for example, is pretty much black and white, the fluidity of hockey dictates that all stats must be taken in context.

From matchsticksandgasoline.com on the Relative Corsi stat, "Robyn Regehr's Corsi REL [in 2010-2011] was -11.4, which means ~11 shots more per 60 minutes of even strength ice time were directed at the Flames' net while he was on the ice then when he wasn't. Now, that may seem really terrible (in fact, it was in the bottom third of the league last year-lumped in with superstars like Deryk Engelland, Cody McCormick and Bryan Allen) but when you take into account a couple of other stats, a number that low is to be expected."

The author, Justin Azevedo, then proceeds to introduce the Quality of Competition stat which numerically incorporates the opposition's players. "QUALCOMP is the measure of how good the opposing players a player is facing are," writes Azevedo. "It's measured by the player's relative plus/minus (plus/minus of when the player is on the ice / [plus/minus when the player is on the ice + plus/minus when the player is off it]) and it's weighted based on TOI versus the player."

And, finally, he closes that section with:

"An Important Note--These two stats, maybe more then any others, require context to ensure they're being read accurately. I don't particularly like these stats by themselves because they can introduce bias based on false assumptions, but I guess anything can do that. Using these two stats in concert, however, is an excellent idea."
 
Back in 2011 when the above piece was written, the Buffalo Sabres under new owner Terry Pegula listed Director of Hockey Analytics and Hockey Analytics Assistant positions. Scott Schranz and Graham Beamish had those titles, respectively.
 
As of today, the Sabres have the word "analytics" as an add-on title for Beamish--Pro Scout/Analytics.
 
Are hey trending way from advanced stats after only three years?
 
It's highly possible as the team has a scouting department that lists 28 pro and amateur scouts and their General Manager came from a scouting background.
 
Sabres GM Tim Murray is an old-school kinda guy. He's been a scout throughout his entire 20-year hockey career and although he acknowledges the "analytics trend," if it were likened to a person he sees on the street, he'd probably nod and continue walking.
 
When asked by the afternoon guys at WGR about analytics Murray was quoted as saying, "I don't know much about analytics.
 
"To me, you go by what you see, you ask the guys upstairs if the numbers correlate, and if they do, then you think you're on the right track, and if they don't, then you have to make a decision: Are the numbers right, which most times they are, or are your eyes right? And that's the tough decision. Because there are players out there you like that don't necessarily have great numbers, and I don't want to get into all my philosophy on all that, but you can get fooled some times, for sure."
 
If the Sabres are trending away from analytics at this time, perhaps it's because it clashes with their views on how they want to lay the foundation and with the type of players they wish to use at this juncture.
 
On the eve of free agency this year, Murray was on WGR's Hockey Hotline and had a bit of a Freudian slip.
 
He was acknowledging that the team needed scorers and doubted that the top-five free agent scorers would be "beating down the door to come to Buffalo."
 
"So what's Plan-B?" he asked rhetorically.
 
"Plan-B is to be hard to play against. To have character guys in the room," Murray answered himself. "We're going to be adding two or three young guys for the next three or four years. Do we just leave it to them to learn how to be good pros? Or do we have people in the room that show them how to become good pros.
 
"That's not the analytics movement pretty guy. That's the reality"
 
Perhaps it's a generational thing or a strict adherence to tradition, but I tend to believe that analytics will soon fade into a tiny compartmentalized area of the hockey department. It will be used, like Murray said, to merely augment, not define as hockey people don't have time for "pretty guy" statistical analysis. They're too busy finding hockey players.

No comments:

Post a Comment