Classic play of yours. Try to find the weakest point of someone's argument and pretend that defines the whole thing. But I'll address your point.
Blueger's hit stats and speed burst numbers have been pretty consistently better than Hoglanders, too.
His PK numbers are overall excellent.
Everything there is wrong, but the usage stuff is just egregious.
And again, it's not just Teddy Blueger. Your numbers are also saying that Nils Hoglander is better than prime Bo Horvat and JT Miller, amongst others.
Blueger hasn't actually hard matched over his time as a Canuck. He's not a particularly good shutdown player and hasn't been deployed like it - he's spent more time playing against 4th lines, which makes sense, because he's a 4th line player. Those minutes were going to Horvat, Miller, Pettersson, Suter etc.
Hoglander hasn't been sheltered. He's played up and down the lineup.
What you have done is taken a model and dismissed it because of your own biases. If analytics is actually the worst thing in hockey, you shoud probably stop watching, because they are only going to get more prominent.
Like, anyone who watches a smattering of Canuck games can see pretty instantly that Blueger plays harder minutes than Hoglander. This is not close. One guy gets 30% zone starts and the other 60. One guy is continually out there in score effects situations defending leads and the other is not. One guy kills penalties. And on and on.
To try and argue otherwise is just using circuitous reasoning that the bad data from these models justifies the bad data.
Again, the central problem is that these models consistently don’t account for usage vs. possession numbers.
You can see this relationship, so strongly, when it comes to Corsi vs. zone starts. The correlation is absolutely incredible and should tell anyone that these things are inexorably linked. When you dump the stats it literally jumps off the page.
But the guys doing these models do studies on the actual value of an individual faceoff, and it tell you that actually this doesn’t matter. And they’re probably right in a vacuum. But what the zone start stats do provide is the *
context* of how a coach trusts a player, and how their overall usage essentially ‘works’ in most cases. And it’s telling you that these models are simply not picking something up, whether it’s valuing the matchups correctly or whether it’s correctly valuing the time of game and situation of icetime or whatever. Because taking these models at face value is telling you that every coach is using nearly every defensive player in the NHL incorrectly and that pretty much every player who starts a bunch in his zone and is trusted by his coach in defensive situations … absolutely sucks. But then magically improves when removed from that situation. It doesn’t make sense. It’s bad math. Parts are missing from the equation. Jon Cooper is not a moron for playing Erik Cernak in defensive situations and thinking he’s a better player than AAAA players like Max Crozier and CA D’Astous.
And it’s absolutely infuriating to see people who don’t understand math parroting bad math at the expense of actually watching the f***ing games and seeing what’s happening.
Per the bolded, I absolutely have not. I've looked at these models for years and can pretty clearly see blatant patterned flaws that renders them useless. I've explained to you what some of these flaws are (the Goldobin Effect as an example) and why these numbers are flawed. What's actually happening is that you're taking someone's model as gospel with knowing how it works or whether it's actually correct, and ignoring all other evidence and pretending that every single person involved in hockey is monstrously stupid. And it just isn't the case.
