This should come as no surprise to you, the faithful PPP reader, but hockey columnist Steve Simmons hates the use of modern analytics in hockey. He hates everything about them. He hates the people who promote them. He hates that they sometimes disagree with his "well trained" hockey eyes. He hates that there's a readily accessible counter to his tired arguments. He hates that he's quickly becoming outpaced by hobbyists.
Over the past few weeks since the Leafs missed the playoffs, as all the stats guys said they would, he's gone on tirade after tirade after tirade after tirade after tirade after tirade after tirade proclaiming that stats can't tell you everything, but his well trained eyes can. (Eyes so well trained that they missed the obituary for Alexander Karpovtsev. Eyes so well trained that they spotted a career ending hip injury to Mats Sundin.) Here's the problem -well just one of the many problems with Simmons' "point" (?) here - nobody is claiming that stats can tell you everything. In fact, the only people who've ever made this statement are anti-analytics morons like Simmons and his ilk.
Statements like those in Steve Simmons' opus titled "Why Hockey's Trendy Advanced Stats are a Numbers Game".
Numbers Game: The use of inappropriate statistics to reflect a desired result; usually misleading.
Right....That's what's going on here. Lots of smart people like lawyers, engineers, accountants, statisticians, college professors, and PhDs are touting the use of analytics in hockey to purposefully mislead people into.....what exactly I'm not sure. Clearly it's not losing money in Vegas, because as Travis Yost points out:
Is it a bad time to remind everyone that Score-Adjusted Fenwick% last twenty is 9-3 picking playoff winners this year, 70-31 since '07?— Travis Yost (@TravisHeHateMe) May 20, 2014
So picking winners based solely on the best team-level possession stat the hobbyists have come up for the last 20 games of the regular season heading into the playoffs will make you right 70% of the time. If you are right 70% of the time in Vegas, you'd be taken out to the desert or on a canoe ride and shot because you'd be robbing the casinos blind and casinos kind of hate that.
Now that we're past the stupid as shit title, let's dive into the body of this crap heap and Fire Joe Morgan the hell out of it in another rousing edition of the Negative Nancy Notebook.
On the night I first began to question advanced statistics in hockey, the stats man who sits a few seats down from me in the press box began regurgitating the game in numbers.
I'm calling bullshit right here. Steve Simmons was questioning and denouncing advanced stats long before Jay McClement became a Maple Leaf in the summer of 2012. So right there, Steve's stupid story is debunked. (Tyler Dellow further debunked it in his post from last night.) So let's just ignore Simmons' obviously fake stories about "sample size" and move on.
The apparently new stats, which aren’t that new, are also historically relevant in baseball. The all-time leaders in WAR — wins above replacement — are Babe Ruth and Cy Young. The all-time leaders in OPS, the batting stat du jour, are Ruth, Ted Williams and Lou Gehrig.
You see those names, you can’t argue back.
The sport is built for and by statistics. Hockey is not so easily determined.
One problem with modern hockey analytics is that the NHL has only provided game data sheets suitable for advanced stats going back to 2007 so we cannot compare historical players like Wayne Gretzky, Mario Lemieux, or Bobby Orr on the exact same playing field. What we can do is look at the 7 years of data since then and see who comes out on top.Looking FF% since 2007, wanna take a guess at who's at or near the top of the list? Just some schlubs like Pavel Datsyuk, Justin Williams, Jonathan Toews, Brian Rafalski, Nicklas Lidstrom, Henrik Zetterberg, Jaromir Jagr, Marian Hossa, Anze Kopitar, Sedin, Zach Parise, and Patrice Bergeron. Are those good players?
And, in a way, the stance to match it with other sports has polarized the game, divided old and new, divided zealot and traditionalist.
Talk about the pot calling the kettle a zealot.
Zealot: a person who is fanatical and uncompromising in pursuit of their religious, political, or other ideals.
On the famed Bill Nye vs Ken Ham debate, the two participants were asked what it would take for them to change their mind or at least question their position regarding evolution. Bill Nye said all it would take was one single piece of counter evidence. Ken Ham said nothing; nothing could change his mind. Nothing can change Steve Simmons' mind when it comes to the value of statistics to hockey.
Hockey is hard to statistically quantify.
Well duh. What Simmons is really saying here is "Statistically quantifying hockey is too difficult a task for me because I lack the knowledge, skills, intelligence, and curiosity to do so. Therefore, these stats are worthless."
If you can’t play without the puck in the NHL, for the most part, you can’t play or won’t play. So how, numerically, do you measure a player when 95% or more of his 45-second spurts is spent without the puck?
You mean like looking at on-ice stats for those players to see how well a team does when a certain player is or isn't on the ice? You mean by looking at WOWY stats to see how players affect their linemates? Of course, one caveat here is that these stats, just like all advanced stats, need to be taken in context. Steve Simmons isn't a proponent of context because context ruins narrative.
/cherry picked examples of lose puck battles and faceoffs leading to opportune goals in big games.
These are game- and series-changing plays: They can’t be defined by any statistic. There is a mistake and a bounce and a battle and a deflection and another bounce and a goal.
This is where Simmons really starts to miss the forests for the trees. These "game- and series-changing plays" happened because of some shots on goal that led to odd bounces. The better possession a team or player has, the more chances for these types of bounces they're going to get. This isn't a difficult concept to grasp, unless your name is Steve Simmons.
The Maple Leafs were among the worst Corsi and Fenwick teams (the best known of the advanced statistics) in the NHL this season. When they collapsed, the stats mavens were almost gleeful. They knew it was coming. They called it.
Good thing the Leafs don't play in the CHL. The CORSI hockey league. They're doing just fine in NHL, though.— steve simmons (@simmonssteve) October 30, 2013
Yeah, but what was their record at the time?
The Leafs were their convenient poster-boy for the changing way to interpret hockey. And an easy target.
Just like Colorado the year before, and Minnesota the year before that....
Speaking of Colorado,
The mavens weren’t quite so accurate in their analysis of the Colorado Avalanche who, like the Leafs, gave up too many shots against and didn’t have the puck enough. But all Colorado did was win and finish ahead of Chicago and St. Louis.
Colorado's still in the playoffs right? They weren't beat in the first round by a team with better regular season possession right? Wait a minute, a team that won more games than it should have based on amazing goaltending which ultimately got bounced in the first round? Where have I heard this before?
The Los Angeles Kings, even before Marian Gaborik, were among the best possession teams in the NHL and yet among the most challenged to score goals. At one point in the season, they scored 16 goals in a 10-game period and followed that up by scoring three goals over six games: That’s 19 goals in almost 20% of the season.
At that time, the team that had the puck the most scored the least.
First off, and this will come as a shocker to you I'm sure, but Steve Simmons has his facts wrong. LA had two 10-game spans with only 16 goals scored: Games 38-47, and Games 53-62. For each span, the next 6 games saw them score far more than 3 goals. Games 48-50 saw LA score 8 goals. Games 63-65 saw them score 9. Not only that, but LA never had a 16 game span with only 19 goals scored. LA's worst 16 game stretch saw them score 24 times. That's not 19.
Also, when you try to make an example of possession not telling a story, maybe don't pick a team that's playing in the Conference Finals. Just a thought. Pick an easy target like New Jersey (goaltending), Nashville (goaltending), Vancouver (goaltending), or Ottawa (goaltending).
The voices of analytics haven’t invented a new game, only a new way to look at it.
No shit Sherlock. It's the same game out there. Thank you for your brilliant insight. The first, and only time I've ever agreed with Steve Simmons. I feel unclean.
Steve Simmons has covered the NHL for more than 30 years and has coached various levels of hockey for more than 20 of those years.
Yeah, but that doesn't mean he's any good at either. It's thinking like this (stupid thinking) that led to the Leafs retaining Carlyle, or even signing him in the first place. This is lazy "appeal to authority" bullshit, if you can even call Simmons an authority. (Note: You cannot.)
There is no statistic to accurately quantify neutral-zone play.
Yes there is. Corey Sznajder is tracking zone entries and exits for every team.
There is no statistic that tell you which wingers gets the puck off the boards and out of their zone and which wingers do not.
Yes there is. mc79 tracks multi-shot shifts.
There is no statistic that defines vision and creativity: That is pure subjectivity.
Yes there is. It's called IPP. (If you say this doesn't quantify creativity; what good is it if it doesn't lead to goals?)
There is no statistic, outside of individual team stats, that measures scoring chances which, again, is a subjective stat.
Yes there is. Lots of people are tracking shot location data to determine what a scoring chance is. Does it align 100% with the subjective definition of a scoring chance? Of course not; it's subjective and varies depending and who you ask. But if you track them all the same way, you get comparable results.
There is no statistic that separates a good dump-in from a bad dump-in. There is a difference, just as there is for a good line change and a bad one.
Yes there is. I already linked to Corey's work on zone entries which classifies dump-ins for change of possession.
There is no statistic that indicates individual ability to win loose puck battles, especially in close games or the last minutes of periods or late in shifts.
Yes there is. Win more puck battles than your opponents and it'll show up in the possession metrics. That's inherent to the belief behind those stats. What Simmons misses is that all of these are statistics that could be tracked if they aren't already. You just need a willingness to do so for example an owner that has billions in revenue and hundreds of millions in profits might want to carve out a slice and track some of those examples for the team, the league, and possibly multiple seasons. It's not exactly rocket science to have someone watch the games and count events and yet it seems beyond the grasp of Steve.
Can't see the forest for the trees: Simply that you have focused on the many details and have failed to see the overall view, impression or key point.
So why is Steve Simmons so caught up in the minutiae of hockey and unable to see that hockey analytics for the most part are about identifying long term trends in the game? Why does he talk about things like board battles, one-on-ones, odd bounces, and loose pucks as being important to winning hockey games (just as everyone else does), but cannot connect the dots between the occurrence of those things and how they are directly related to overall possession trends? Why does he talk about stats being useless because hockey is such a random game, and yet poke fun at the stats guys when they talk about the randomness of small sample sizes? Why? Because quite simply, Steve Simmons is dumber than you.
Steve Simmons can be reached via Twitter @simmonssteve, via email at firstname.lastname@example.org, or via text message at 416-723-6379.