Proponents of advanced stats want to apply their logic to all aspects of hockey. However, applying percentages to the Colorado Avalanche is not useful.
“What matters is who wins the game.” ~Joe Sakic
The Colorado Avalanche are infamous for being very low on the totem pole according to advanced stats. I’m not going to pretend to understand the mathematics involved in Corsi and Fenway — I know it’s Fenwick, stay with me. However, I know possession stats and shots against stats are very bad for the Colorado Avalanche.
It’s hard to put an exact number on possession. You can look at how often a team is in its own defensive zone versus the offensive zone. You can also use shots for versus shots against as a means of possession.
And then you can watch in confusion as one of two things happen. You can watch the Colorado Avalanche have excellent numbers and lose anyway as they did against the Vancouver Canucks. Or, you can watch them have terrible numbers and win anyway as they did against the Montreal Canadiens.
These aren’t one-offs. Again and again, the Colorado Avalanche live out that reality. Good stats = losing, bad stats = winning.
And yet, advanced stats proponents harp again and again on how bad Colorado’s stats are. If the team happens to have a bad stats = losing game, they instantly jump on the one leading to the other. “Colorado is a poor possession team — that’s why they lost!”
If the Colorado Avalanche have the more typical bad stats = winning game, said analytics proponents shake their heads. “This success is unsustainable.”
If the Colorado Avalanche have the kind of new good stats = losing game, the analytics proponents seem to just shrug.
Head coach Patrick Roy has pointed again and again to the reality that Colorado loses when they have good statistics. Although, there was a new hybrid game against the Florida Panthers the other night. For two periods the Colorado Avalanche played a good game and seemed like they were going to win! Then, they played a bad period and… won anyway!
Goalie Calvin Pickard actually broke his coach’s record of most saves in a period with 24. He didn’t let a single one in all period. But the fact that he faced 24 shots in a single period tells you how badly the game broke down for the Colorado Avalanche.
Like I said, though, prior to that the Colorado Avalanche were already winning despite playing well. (I didn’t mistype that.)
I think proponents of analytics like fancy stats because they’re completely black and white. Puck possession = shots = winning. If you’re not winning, it’s because the first two parts of the equation aren’t in place.
Ok, fine. What do advanced stats say a team needs to do to possess the puck more? Well… get the puck, maybe? Ah, I know — get the kind of players who are better at puck possession.
The Colorado Avalanche catch a lot of flak for letting Paul Stastny go and for trading Ryan O’Reilly. Both were good puck possession guys. (Colorado didn’t get past the first round of the 2014 playoffs even with those players, but never mind that — they’re Corsi and Fenway, er, Fenwick were bad for different reasons.)
Analytics proponents love to use fancy stats to prove a point about a player. They’ll point to a player’s Corsi-for or Corsi-against as proof to how valuable that player is to the team.
My personal favorite is when they point to those stats as if they’re showing them to Avalanche GM Joe Sakic and head coach Patrick Roy. “See. This player you’re using isn’t any good. You shouldn’t be using this player.”
Think about your life for a second. Think about all the numbers that can be associated with it. Are you the sum total of your height and weight — is that your worth? How about your bank account — does that say everything about your life? Consider your work. Can an outsider use nothing but mathematics to evaluate your worth to the company? In some cases, like sales, yes. How about a teacher? A surgeon? An artist? Math might tell you something about their work, but not everything.
So goes it with a player. You could look at a player like goalie Semyon Varlamov who seems to get better the more shots he faces. That flies in the face of logic — more shots = more goals, right? Ergo, fewer shots should equal fewer goals.
Look at Matt Duchene versus Nathan MacKinnon. MacKinnon has 30% more shots than Duchene. Ironically, Duchene has almost 30% more goals than MacKinnon.
Coach Roy once laughed on 104.3 The Fan that if players were trying to manipulate the fancy stats, they’d just start shooting from anywhere — “the blue line, the zamboni door, the bar across the street from the arena.”
I’m not saying MacKinnon’s trying to manipulate the stats. Rather, he seems to be employing a “shoot first” mentality. Duchene prefers to pick his spots. Duchene’s first on the team in scoring and MacKinnon’s second, so their individual approaches seem to be working for them.
That brings me to my main point of what I think is more important than advanced stats in a game and in a player — the person. Each Colorado Avalanche player is the sum of his parts, including statistics but also talent, psychology and fitness. Indeed, I daresay talent, psychology and fitness produce a player’s Corsi-for/against and whatever Fen…wick tells us.
The great minds of hockey, such as Sakic and Roy, can tell you everything you need to know about each player. Place a graph in front of him and indicate levels. I daresay either Sakic or Roy can show with great accuracy where each player is on the graph without looking at a single advanced stat.
From there, he can also tell you more, such as the fact that Tyson Barrie’s defense flounders when he gets tired, but he’s still the player you want on the ice when the team’s down by one or two late in the third.
Advanced stats have their place. They can help people who aren’t hockey masterminds understand the game better (if you happen to be good at math — they tell me next to nothing). They can help explain some aspects of the game.
However, I prefer a more gestalt approach to hockey, and it seems that’s Sakic’s and Roy’s approach as well. It’s true that they hired an analytics expert last summer. However, neither seems to be relying on said expert.
Instead, they’re considering their players and their division and conference. They’re relying on the hockey knowledge that allowed both of them to win Stanley Cups and get into the Hall of Fame to evaluate players, division, conference.
From there, they’re plugging the players they’ve got into a system designed to complement them. They’re changing the lineup and the lines as new information presents itself — even within the game, as we’ve seen with coach Roy.
I like that approach. I like knowing that Mikkel Boedker is a speedster who can keep up with Nathan MacKinnon and open the scoring up. I like knowing that Matt Duchene enjoys playing with a veteran like Jarome Iginla who yaps about the game almost as much as he does.
Next: Goalie Controversy for the Avs
In the end, coach Roy isn’t a math guy. He caught flak a while back because he seemed to get the advanced stats system Fenwick confused with the park Fenway. I propose the slip was as deliberate as mine in this post were — perhaps more so. Coach Roy was indicating how superfluous analytics are to his part in the game.
At the end of the day, as Sakic points out, it’s about who wins the game. Who cares how they got there?