The VICE Basics to Hockey Analytics

This story originally appeared on VICE Sports Canada.

"You can lead a horse to fancy stats, but you can't make him use them, understand them or accept them." – ancient Canadian proverb

It's 2015, a time when we as a society have so much new technology and information that wasn't available as recently as 10 years ago. Whether it's advancements in the medical community, the ability to watch NHL games on a mobile device (on a two-minute delay) or the fact that someone combined a croissant and a donut, it's a glorious time to be alive.

Yet there remains some resistance to analytics, shot metrics and fancy stats in some NHL circles.

Why? No one likes change. No one wants to do things differently than they have all their lives. No one likes the idea of putting in time to learn something new. For some, new ideas can be scary, even if the "new" idea is simply counting shot attempt totals.

This is trampled ground, to be sure, but maybe this piece is what prevents someone from making another crack about Fenwick after Anton Stralman makes one bad play in one game in March. Here are some basics about fancy stats that could help you avoid a nine-hour Twitter fight that's already been had 8,000 times.

Why are you doing this to yourself?

Self-loathing? The desire to educate? The fertile ground for jokes? A little bit of all three, I suppose.

What do you think is the biggest reason why advanced or fancy stats have met so much resistance?

As with anything in 2015, there's nothing more important than branding. Curating unique pos-men engagements across digital channels and social platforms are crucial for level-setting and gaining brand awareness.

Labeling anything "fancy" or "advanced" will always result in people recoiling in fear. Would you like fancy mustard or regular mustard? Do you want to enroll in an advanced yoga class or a yoga class? If something is new to you, the odds of you engaging with that thing aren't very good if you are intimidated by it.

But here's the trick: Advanced stats or fancy stats aren't all that advanced or fancy. In their most basic form, they are quite simple.

So what's the most basic advanced/fancy stats thing I should know?

The terms everyone should know are Corsi and Fenwick.

Corsi is the number you get when tabulating a team's or player's shot attempts. Fenwick is the same exact thing, only you do not count the shot attempts that are blocked.

That's it?

That's it.

Phew, great, that's extremely easy…

Well, not entirely it.

What? Come on.

No, just stay with me. The important thing to know about Corsi and Fenwick are the only shots we are talking about are the ones at five-on-five.

OK, that's easy enough. So if Phil Kessel has 11 shot attempts at five-on-five, his Corsi is 11?

Oh, sorry, there's another part of this to understand. Corsi and Fenwick are five-on-five numbers in every sense of the word. Yes, Kessel's shot attempts are part of that, but so are the shot attempts of his four teammates on the ice at the same time as him. That's why you hear the phrase "on-ice shot attempts" sometimes. So if he has 11 and the other four teammates on the ice with him have 18, then Kessel was on the ice for 29 Corsi events.

Events?

Let's not nitpick extraneous words in the English language. Call them Corsis or Corsi danglefeffers. I don't care.

So then Kessel has 29 Corsi events in that game. Why do I always see those numbers in the form of percentages?

Because Corsi and Fenwick are determined by how well you do against the other five guys on the ice. So let's say Kessel was on the ice for 29 five-on-five shot attempts for his team and 18 five-on-five shot attempts for the other team. That means he was on the ice for 29 of 47 shot attempts, which translates to a Corsi of 61.7 percent.

Here's the part that bugs me. You say Kessel has a 61.7 percent Corsi in that scenario. But how do I know how involved he was in that number? How do I know it wasn't the other four guys on the ice that were driving that number? Doesn't pegging a guy with an individual number based on a five-man performance seem unfair?

That's a great question. It seems like you're starting to follow this. This may seem somewhat advanced, and this is as advanced as we will get here, but this is where WOWY comes into play. It stands for With Or Without You, and if your next question is whether this idea was invented by Bono and The Edge, the answer is yes. U2 are the fathers of hockey analytics.

WOWY shows an individual's statistics with or without certain players on the ice. So while Kessel was at 61.7 percent overall for that game, his numbers may be slightly higher or lower when you look at a single person with whom he was playing in that game. These sorts of numbers aren't all that helpful when looking at a single game, but over the course of a season, you can learn which players are the ones driving possession and the ones acting as anchors around the necks of teammates.

That's all well and good but I'm always hearing how these numbers have "predictive" qualities? How so?

What Corsi and Fenwick do is give you an idea of how much a team is possessing the puck, hence the reason why you hear "possession" all the time. If you have the puck more than the other team, in theory, you are giving yourself a better chance to win. There is a crazy amount of luck in hockey, so if you are putting more pucks toward the net than the other team, you are giving yourself more chances to score and for luck to help you.

Yeah, that wrist shot from the blue line that bounced off a stick, then a skate, then a knee pad, then the post, then the goaltender's butt and into the net is a lucky goal, but it's luck that's only possible when you have the puck. While that goal isn't repeatable, the skills required to possess the puck are.

The way I like to think of it is in relation to poker. Poker is a skill game where luck plays a factor in outcomes, just like hockey. Occasionally, luck will go against you—sometimes your opponent will catch that one card that beats you and sometimes your opponent will score four goals despite a low volume of shot attempts compared to your higher volume.

If you are consistently above 50 percent in Corsi or Fenwick over 82 games and consistently getting your money into the pot with a better-than-50-percent-chance to win the hand, the idea is you will win more than you will lose over a long enough timeline. The repeatable skill of possessing the puck and directing it toward the other team's net will result in more wins than losses; the opposite should be true of teams that don't have the puck as often.

Somehow, none of that applied to the Kings (54.7) and Flames (45.8) last season. It's a good predictive tool, but it's not perfect. That's usually enough for stat deniers to slam it. It's sort of like hating vaccines because they are only 95 percent effective.

So should I fight with anyone who mocks fancy stats?

No.

No, really, I can have an eight-hour argument on Twitter about it.

Please don't.

The people will want to hear it.

I'm begging you to just let it go. The stat wars are over. Stats have won. Almost every NHL team has a stats person or a stats department. At this point, you'd be fighting with one of those soldiers that haven't gotten the word that the peace treaty has been signed and the territories have been divided among the statisticians. He's lobbing malfunctioning grenades at ghosts.

One last thing: Luck. I always hear about it. Is there a way to quantify it?

Yes. You should look to PDO to get a feel for which teams are playing over their heads or are primed for a bounce back.

OK, cool. What does PDO stand for?

Pushes Doors Open.

Really?

Pukes Dumbledore Ostriches.

*Is this serious?*

Actually, PDO is a lot like American politicians in that it doesn't really stand for anything. It could stand for Prosperity Dynamic Optics but it doesn't. It's just PDO. That's what U2 wanted, so we go with that.

So what is it?

Again, it's all about five-on-five play. The idea is a team's shooting percentage and save percentage should add up to 100. This is why you hear the term "sustainable" a lot.

So if a team's or player's PDO is well above 100 at 35 games into the season, the chances are pretty good it will come down, especially if they are sitting below 50 percent in shot attempts. The same goes for a team that's well below 100 in PDO. What those teams or players are doing is very likely unsustainable and things will turn one way or the other. If you see a team winning a lot of games with a shooting percentage near 10, there's very likely a drop coming.

Last season, according to stats.hockeyanalysis.com, the 45-win Flames had a PDO of 101.0 and a Fenwick of 48.0 percent. This season, the Flames have a PDO of 92.9 and a Fenwick of 45.6 percent. Not surprisingly, last season's PDO darlings are 2-7-1 through five games, as their luck isn't quite what it was in 2014-15, but their possession numbers are unchanged.

Are we done here then?

Yes.

What about Fenwick-close and how it's better than Fenwick because it mitigates score effects? What about zone starts and how they can inflate or deflate a player's possession numbers? You didn't even touch on how quality of competition can also play a role in possession numbers. What about points per 60 minutes and the value of that number?

Look, this was for beginners. You could have said something earlier if you knew all this already.