Five things: On trolling and analytics

Everyone is up in arms about something Steve “What ever happened to Alexander Karpovtsev” Simmons wrote on Monday, and justifiably so. It’s something so bad and disingenuous — and it won’t be linked here — that he should be embarrassed to have attached his name to it. As with most Steve Simmons articles these days, it is obsessed with the idea that there are people who think they know more than he does, and how he has no ability whatsoever to handle the fact that they might be right.

The thing with this is that it’s blatant trolling of a group (the hockey analytics community) that is regularly looked down upon by people in positions of power within the hockey media. Now, when I talked about this steaming heap of garbage on Twitter Monday night, I once again put forth my common assertion that I do not troll, and this was generally laughed at as being preposterous.

But the thing is this, and it’s an important distinction between what I do and what Simmons does: Trolling is the taking on and advocating for a point of view that one does not necessarily believe, or even amplifying a held belief, to make people angry and generate page-views. That’s what I’m accused of all the time. I have never, though, written something I didn’t believe without exaggeration — apart from the anti-Canada stuff, but only idiots don’t see that stuff as being pretty obviously done for the sake of humor — nor have I thought that something I wrote would get a lot of clicks. Don’t read this. I don’t care at all. Don’t comment. I would actually prefer you didn’t. (The proprietors of this site, however, likely do, so sorry about that.)

I believe that the best way to rebuild a team is to blow it to smithereens. I believe Sean Monahan should have spent the year in junior no matter how well he did at the NHL level. I believe Brian Burke is incorrect about where this sport is going and is therefore not the man to lead the Flames into the future. I believe it’s wrong for the organization to “sell hope” in a crass attempt to pump ticket sales while potentially impeding future progress toward meaningful competitiveness. I believe in all that and I’ll swear on a stack of bibles that I do. You might not agree with me, and it’s fine if you don’t (as long as you don’t mind being wrong, because you are), but that doesn’t mean I’m trolling you. It means we have opposing viewpoints. That’s life.

Moreover, and more germane to this debate, I believe in the power of hockey analytics to predict the general outcomes in the league, because we have years of evidence that something as simple as raw even-strength Corsi percentage works more often than it does not. To say that it doesn’t work, as Simmons regularly does, is to simply ignore evidence and then act like an asshole about it in exchange for money. That’s the definition of trolling.

One of the things that was brought up in the immediate wake of the publication of Simmons’ trite, smug nonsense is that this “debate” about the efficacy of hockey analytics has grown tiresome. It’s true to some extent.

This “debate” is tiresome in the same way the climate change or evolution “debates” are tiresome. There are always going to be people who hold steadfastly onto their beliefs — or continue peddling what they know to be snake oil in exchange for profit — no matter how much evidence you present them. The great new HBO show “Last Week Tonight with John Oliver” recently covered this topic in an interesting way, holding a debate between Bill Nye, and a representative 97 percent of climate scientists who say climate change is real, and the representative 3 percent who say it’s not, or it’s too early to tell.

The numbers scream that Corsi is an excellent (but not perfect) tool to estimate future results at the NHL level, and maybe people are right that it’s a debate that goes around in circles and gets everyone nowhere. But that same reality doesn’t prevent Bill Nye from going on cable news shows and too-politely shouting down luddite idiots.

That’s all the evidence you need that Corsi works. Depending on how you use it and narrow down the range down, it’s an accurate predictor of playoff results about 70 percent of the time. Point to the outliers all you want — and those who don’t want The Nerds to be right do this constantly — but you’re pointing to the smaller number, and only the deeply stupid know that but don’t care.

Which brings us to the actual substance of what Steve Simmons wrote. He led off his column with anecdotes about how Mikhail Grabovski was better than Jay McClement in such-and-such a game according to corsi. When he asked a “stats man” (who may or may not have been James Mirtle) why he and another colleague both felt McClement was better, the answer from this nameless man was, “Sample size.” This was allegedly the same answer given during Game 1 of the first round last season, when the Leafs were buried by the Bruins, and James van Riemsdyk was supposedly the best Leaf on the ice according to both “the numbers” and “the eye test.”

Tyler Dellow looked it up and found that the latter claim (that van Riemsdyk was the best Leaf according to corsi) was a complete falsehood. We can also go ahead and doubt the truthfulness of the quotes in either case, because no strawman stats man would answer “sample size” to childlike questions such as these. It’s not an answer that actually applies to the query at all; a better answer would be “Because Grabovski/van Riemsdyk is a better hockey player than McClement/most other people on the Leafs.”

Later in the column, Simmons attempted to disprove the usefulness of corsi by saying that if the Kings were so dominant at even strength (and boy, are they ever), then why was there a stretch of 16 games in which they scored just 19 goals? The answer to this question, according to SkinnyPhish, is that there wasn’t. The fewest goals the Kings scored over any 16-game span this season was 24. That’s still not a lot, but it’s better. Let’s put it this way: East coast bias or not, the hockey world would have been talking at length about the Kings’ offense if they’d scored 19 goals in 16 games.

This also ignores the fact that Corsi doesn’t have anything to do with shooting percentage, or that even if they had scored 19 in 16, which again they did not, it didn’t really hurt them too badly because they made the playoffs with ease and in fact won 100 damn points this season. It further ignores that the Kings are in the Western Conference Final for the third (THIRD!!!!!!!) straight season, so they must be doing something right.

So that’s at least two things Simmons was objectively incorrect —and potentially lied — about, and missed the point about besides. He may also have been wrong or missed the point of what his “stats man” said about the Leafs, but these are personal anecdotes that, fortunately for him, cannot be fact-checked with a web browser open to Hockey-Reference and Extra Skater.

Again, the problem with this “debate” is that there’s nothing to do with it. Both sides are dug in. The people who support analytics aren’t going to change their minds and go back to the other side, because they know they’re right and have mountains of proof on their side, and can’t be swayed back to the side of wrongness. That’s really not how intellectual thought progresses.

The people who are skeptical or outright dismissive of analytics will either be won over by the growing evidence (this is what happened to me), or go to their stupid graves swearing they’re right while that evidence and the growing numbers of people who accept them as more or less true know better. Self-satisfied bones in a box on their way to being literal fossils, after spending their years being figurative dinosaurs on the subject.

In the meantime, the sport will continue to pass them by, and their predictions will continue to be less accurate, and they’ll feel more and more surrounded by Nerds With Calculators And Spreadsheets Who Don’t Watch The Games. Because that’s just how the games will be watched by everyone.

When they unveiled the new AHL team the banners behind the guys spelled “CACA.” That was funny.