Last week, the hockey world was set ablaze by a (since-denied) story that the league was quietly planning a four-team expansion within the next two years, with the new franchises earmarked for Seattle, Las Vegas, Quebec City, and Toronto B.
The story seemed fishy from the outset, since the NHL has consistently ruled out expansion as an option in recent years, let alone a dramatic four-team expansion that would make it the largest sports league in North America.
That didn’t stop people from trying to analyze the story from all angles, mostly based on opinion and conjecture. And any time there’s a story that begs for deeper analysis, you can bet that Nate Silver’s ESPN-owned FiveThirtyEight will be there, mashing a couple of arbitrary stats into a spreadsheet and treating the results as Gospel truth.
Since storming onto the scene as a political projection tool, FiveThirtyEight has gradually started creeping into other areas - sports, pop culture, world events, fucking burrito rankings - by trying to apply stats and suss out deeper truths that can only be revealed through an objective, statistical approach. It’s a pretty noble endeavor.
The problem all too often with FiveThirtyEight, though, is when they bite off more than they can chew with a project, and then stubbornly stick to their analysis even when it doesn’t pass the sniff test, or when their so-called “objective” results are based on a series of purely subjective, arbitrary assumptions.
The area where FiveThirtyEight traditionally shines is in political predictions. There, not only are stats wonks blessed with an all-you-can-eat buffet of free, up-to-date statistics and polls to play with, but the actual outcomes of elections are much more finite. You can feed a bunch of Quinnipiac polls into a regression model, and feel pretty good predicting a Republican victory in Oklahoma’s 3rd district.
It’s a lot harder, though, to make definitive statistical projections in more volatile areas, like pop culture, or sports - places where there are too many variables, or the variables are less reliable, or there are far more possible outcomes to consider.
Take sports as an example: using their handpicked criteria, FiveThirtyEight, like every other prediction model, was all but certain Spain and England would advance in their elaborate World Cup projections. They posted 65% odds that Brazil would beat Germany (Germany just barely edged Brazil, 7-1), refusing to alter their formula as the tournament progressed and new information became available, but instead stubbornly sticking to their guns that Brazil was statistically the best team in the tournament. There’s nothing wrong with trying to use stats to make projections in more volatile fields, like sports and arts and culture - it’s just that you should know when to fold up the model you created and head back to the drawing board.
On the heels of the NHL expansion news, FiveThirtyEight went to work, trying to apply their patented number-crunching approach to see if this was a good idea, building off of earlier work they’d done on NHL markets.
In their article, boldly titled “Half of the NHL’s Rumored Expansion Cities Don’t Make Sense”, they make this claim (bolding mine):
Our research showed that, in addition to Quebec City and a second Toronto franchise, the Canadian cities of Kingston, Halifax and perhaps even Moncton, Sherbrooke or Sudbury could each reasonably hope to support a team. From the standpoint of fan avidity, all were more attractive markets than Seattle — not to mention Las Vegas, which was sandwiched between Milwaukee and Kansas City, Missouri, as the least hockey-mad of the potential expansion sites we examined.
That’s a real paragraph that appeared in a real publication owned by Disney-ESPN.
It’s breathtakingly wrong about everything.
How did FiveThirtyEight get to this insane conclusion? That fucking Sudbury, Ontario - a sleepy nickel-mining city of 100,000 in Northern Ontario, best known for having a large smokestack and a song about a bunch of destitute losers getting drunk and playing bingo - would be better-suited to have an NHL team than Seattle, with its metro population of 3.6 million, massive corporate base of Fortune 500 companies, and an actual potential owner?
1. They only used a single metric for their analysis - and it’s solely based on Google searches for the word “NHL”. The crux of the whole article, where they boldly and without hesitation say which cities could and couldn’t support a team, is based on some earlier research Nate Silver did in 2011.
Their sole determining factor back then for their “NHL Avidity” chart was the number of people in a given city who searched for the word “NHL” on Google.
That’s it. Nate took the population of a city, saw what percentage of people had Googled the term “NHL” in the past year, then used those Google Trends search results as his baseline number of NHL fans in a given market.
If you’d just presented this in a second-year statistics class, the professor would be trying to spout out so many questions at once she’d get tongue-tied:
- Why only use search results, and not any other kind of supporting stats or opinion polls?
- Why only look at the standalone term “NHL”, and not other terms, like specific NHL teams or players, or just the word “hockey”?
- Wouldn’t an existing NHL fan know to go to NHL.com or a specific hockey site, instead of having to type the word “NHL” into a search engine?
- Did you only use Google stats because Google Trends is really easy to use and, more importantly, free? See me in my office Nate, I’ve got a lot more questions.
2. There was nothing else to support the analysis, other than Google hits. Just off the top of my head: The local economy. Population. The arena situation. Infrastructure. Local media. Corporate base. Ownership. Advertisers. Demographics. Geography. Conflicting fanbases. Proximity to existing markets.
These are things I’d expect in a feasibility study. If you’re coming at me with a table you cribbed from a search engine, don’t just authoritatively say “our research shows that” like you’re spitting out hot truth.
3. Maybe worst of all, the “markets” they used in Google Trends are hilariously inconsistent. Sudbury only has a population of around 100,000, making it comically small for an NHL team. FiveThirtyEight, though, grouped “Sudbury-Thunder Bay” as a single market for their analysis, pushing the market’s population to a full 523,000 people.
Sudbury, Ontario is a 12-hour, 1,000km drive from Thunder Bay, Ontario.
New York City is closer to Florida than Sudbury is to Thunder Bay.
These stats somehow created a single market that comprises all of Northern Ontario, a land area larger than all of Texas. I know some teams have big footprints, but the feasibility study’s trying to pinpoint the local population who could realistically go to a game. Hey Thunder Bay fans, come watch a game of your hometown Sudbury Beaver-Trappers - only a two-day drive away!
“Kingston-East Central Ontario”, too, would seemingly cover a gigantic area of pretty much everything between Toronto and Ottawa, if they’re cheating the market population up to 725,000 (Kingston itself sits around 125,000.) They’re probably not going to draw people to the arena who live four hours away. Also, in an area where everyone is already a lifelong fan of an existing team.
Again, I feel like I should remind you that “Sudbury could [...] reasonably hope to support a team” was part of a published story on a site that ESPN purchased from the New York Times for an undisclosed number of millions of dollars.
It’s easy to crunch some Google stats from an office in New York and tell us which Canadian cities can support an NHL team. Let me tell you, as a Canadian, about the five cities FiveThirtyEight says are more viable NHL markets than the wealthy, metropolitan, sports-crazy, arena-having city of Seattle. It’s not even that I’m some Seattle booster or anything - just that there’s just no goddamn way in hell any of these five cities get an NHL team in my lifetime.
SUDBURY is a fucking hole. The people are friendly and there’s a nice science museum, but we’re talking about a barren, far-away, post-industrial mining town of 100,000 with no corporate base. The only areas where Sudbury outproduces Seattle are in nickel exports, meth exports, and operating-an-ATV-while-drunk convictions.
It’s cleaned up a lot, but for decades, Sudbury was so polluted from the mining industry and its massive, smoke-belching smokestacks that it looked like a newly-discovered level of Dante’s Inferno. No grass, no trees - just rocky outcroppings, black soot, frigid winters, and unemployed alcoholics. I had a friend from Sudbury, she said that when her grandparents first arrived there as immigrants from Italy, her grandmother broke down and started crying when the train stopped and they first saw their new home, the pollution-choked hellscape of Sudbury. There she was - having escaped Mussolini - and she was weeping at how awful Sudbury looked. This is not an NHL city.
KINGSTON is known for two things: Kingston Penitentiary, the largest prison in Eastern Canada, and Queen’s University, a school for preppy dipshits. One is a place where people live in squalor for the mistakes they’ve made in their lives, and the other is Kingston Penitentiary.
Kingston is full of trashy, Ricky & Julian-type hosers. I went to Kingston once, went to this rundown townie bar, and saw a girl making out with a stranger, followed by that girl’s sister making out with the same stranger, followed by - I shit you not - the two sisters making out with each other. The next day, I was hanging out with a local who asked me “hey, wanna go down to the Dollar Store to shoplift some shit to make a bong out of?” There are three trashy things about that sentence. This is not an NHL city.
SHERBROOKE is close enough to Montreal that I guarantee every single Sherbrooker is already a Habs fan. And this wouldn’t just be friendly competition - Habs fans derive their entire self-worth from the fact that they identify with a hockey team. Some Bloc Québecois Parliament backbencher would probably try to force the government to block competition to the Habs, on the grounds that it was an affront to French-Canadian heritage, or something.
MONCTON is a dumpy, forgettable town in New Brunswick. The biggest tourist destination in Moncton is the mall. A while back there was a lunatic who went on a shooting spree, and when I saw it reported on CNN, they had to clarify “Moncton, which is a city in the province of New Brunswick, which is in the eastern part of Canada.” When you say the city name and province name to an American and they still don’t even know what country that’s in, you do not have the name recognition to be an NHL city.
HALIFAX is nice, though. They’re still not getting a fucking NHL team anytime soon.
- Quebec City’s good to go. Expand there, and Seattle. Or move the Coyotes, whatever.
- Never listen to Nate Silver.