Colorado Avalanche Analytics: Hockey is not Baseball


The Colorado Avalanche illustrate how hockey is not baseball — making advanced stats only partially effective in explaining the sport.

“The part I don’t like about Corsi is you could shoot from the red line or […] from a terrible angle and your Corsi will look good.” ~Patrick Roy

The Colorado Avalanche are my only team in my only sport. I kind of like the Ottawa Senators and soccer in the same way I kind of like Blue Moon beer — ok in small doses, but not something about which I’m passionate.

To me, baseball is like Pilsner Urquell — definitely not my taste. However, I attended my first Colorado Rockies game with fellow Colorado Avalanche fans, a couple of whom actually love baseball.

They were kind enough to explain parts of the game to me — the parts I didn’t learn in elementary gym class. At one point, they had just gotten done telling me a certain player was really good then told me the numbers next to players’ names on the jumbo-tron indicated something good (batting average). The higher the number, the better.

At that point, I observed the player’s number was around .225, which was lower than a lot of other players. My friend said, “He’s been injured. The number changes throughout the game — his will go higher.”

I puzzled over that until I related it to goalie save percentages, a comparison I floated out. My friends were unconvinced, so I tried forward shooting percentages — that makes more sense because it’s offense in both sports.

However, I think batting average is not really like save percentage or shooting percentage except that they’re both averages. No goalie is ever going to be allowed to get down to a .225 save percentage before getting yanked. And very few skaters shoot with 22.5% accuracy — former Colorado Avalanche Alex Tanguay led the NHL a couple years ago with a similar percentage.

In that vein, hockey and baseball aren’t the same except they’re both team sports played with sticks.

Everyone says baseball is a “thinking man’s game,” I’m guessing because you have a lot of time to think while it’s going on. In that case, hockey is a “watching woman’s game” because that’s what you have time to do during a game. Hockey is fast.

And all of that is just a long way to point out that applying the same kind of observation to hockey as you do to baseball makes no sense.

I’ll be the first to admit I don’t understand analytics — I understand baseball better, and my friends had to explain to me what a bull pen was! (I still think the waiting pitchers should look more interested in the game, like backup goalies do.) However, I’ve fooled around on sites long enough to recognize advanced stats are just one limited way to look at teams and players.

It’s difficult to qualify how good a player is. With forwards you can look at scoring, and you can look at save percentages for goalies. Defensmen get the big screw because it’s really hard to put numbers to defense.

However, advanced stats proponents like to do just that. They look at things such as shots per game and translate that into possession stats. They try to quantify how good a player is by how well he and others do when he’s on the ice. That’s fine, even interesting.

However, it’s only one aspect of the game, and one that’s just as subjective as the eye test when it comes to hockey. Because hockey isn’t baseball — you don’t have one player going up against another à la a batter and pitcher. Rather you have a player who passes to another player while being checked and a goalie being screened by an opponent. There are just too many variables in every single shift on the ice — every single play barring the shootout is a team effort.

You might have a little bit of a point when you use advanced stats to evaluate a team — such as those who are always calling the Colorado Avalanche bad because of their possession numbers. Have the Colorado Avalanche had two losing seasons? Yes. Do the Colorado Avalanche allow more shots than they take. Yes. Does that indicate bad possession numbers? Possibly. Is that the reason they had two bad seasons?

Poor possession numbers can be a part of the reason they failed to make the playoffs the last two seasons in a row. Does it explain their sudden collapse in the last nine games of the 2015-16 season? No — why would it all suddenly catch up with them? Either they collapsed because something happened in the locker room or because their two best players — Matt Duchene and Nathan MacKinnon, went down with injuries at the same time. Maybe something else went down as well.

My major problem with advanced stats is how they’re used as a blunt weapon. Either you believe in advanced stats, or their proponents will bludgeon you with them. You’re looking at hockey with old-fashioned eyes if you don’t use advanced stats.

What’s more, anyone with a head for numbers can be a wizard at the game. A lawyer who can’t even skate — such as one of the founders of advanced stats, Tyler Dellow — can supposedly understand the game better than anyone. Including those who got into the Hockey Hall of Fame playing the game.

Such math geniuses can use those stats to show how player A is better than player B, and Hall of Famers who disagree with them are foolish. They then whip up their readers — because advanced stats started as a tool for bloggers — to call out said Hall of Famers.

Perhaps by now you realize I’m referencing two of the biggest opponents to advances stats, Colorado Avalanche head coach Patrick Roy and GM Joe Sakic. I’m not going to bother giving you their hockey player credentials — suffice it to say they’ve won enough awards and trophies to have made them shoe-ins for the Hockey Hall of Fame and to ensure their numbers were retired by Colorado. (And the Montreal Canadiens in Roy’s case.)

How is it that two smart men with vast hockey knowledge can suddenly be so wrong about hockey? How can someone who can’t even ice skate understand hockey better?

The answer is he can’t. You can’t. I can’t. We can use advanced stats to get a shallow view of a fast and dynamic game. A game that men like Sakic and Roy see with depth and complexity.

I hope one of the advanced stats bloggers — perhaps Dellow himself, who now works for the Edmonton Oilers (cue laugh track) — gets to meet Roy or Sakic and ask him a very pointed question. I hope he shows the Hall of Famer a graph with dots but no players’ names that somehow indicates how good said players are.

Next: Tiers of Fandom in Avs Nation

Because once Roy and Sakic understand what each point on the graph means, I’m sure they can name each player in that position. Because they know their hockey, and they know their players.

I’m even willing to bet on it — I bet their first Stanley Cup together on it.

*Editor’s Note: The Patrick Roy quote above is one of his famous disparagements about advanced stats.