[Itse]

Originally Posted by driveway

Fighting is like winning faceoffs.

Over the course of a season, winning faceoffs doesn't matter. Faceoff winning percentage has almost no correlation with winning games, even though to the eye it seems like being a good faceoff team would have a huge impact.

Likewise, the majority of fights are not going to have any impact on the game, either positive or negative.

But every now and then, there is an instance where a specific faceoff win or a specific fight can have a significant impact on the outcome of the game. The problem is that this can only be determined after the fact. There is no way going into a fight or faceoff to know if it will be vital or simply another meaningless occurrence, and there is nothing inherent in the event itself which will lead to it being vital or meaningless.

A bad pro-fighting example actually. Faceoffs are more an example that as long as connections exist, they can usually be found even when there's a lot of noise and randomness involved.


A tangent on faceoffs:

Spoiler!

Statistics suggest a good faceoff team in the NHL will get on average about 2-3 more points over the season than a bad one, all other things being equal. If somebody says faceoffs are not important they probably have a poor grasp of both statistics and NHL standings. (One reason why I don't think much of the "advanced statistics" crowd is that this claim seems to be rather common in those circles.)

To elaborate: if we accept your (IMO quite common and very reasonable) assumption that faceoffs produce extra goals (and thus points) somewhat randomly, it opens up a possibility for quite a bit of variance in how many extra points faceoffs can produce over the season. For example 6 extra points for might happen every season for some team. Even though we can't pinpoint after the season who got extra points for faceoffs and how much, and even though we can't be 100% sure that the correlation is significant, that doesn't matter much. As long as statistics suggest that there's a reasonable chance that it's significant, it's worth caring about. Especially since eyeballing the game goes with the statistics here.

Essentially, stats say we probably should not ignore FO%.

If you ask me, the corsi crowd tends to snub FO% because it doesn't lend itself to snubbery. Nothing discovered so far goes against commonly accepted truths, you can't use it to explain success or failure after the fact, nor is it very good as a prediction tool.

Now, it's perfectly reasonable to suggest that some fights matter even if most don't, I accept this. However, if a notable amount of fights mattered, that should show up somehow, just like with faceoffs.

Which is why I'm of the opinion that without further evidence the only rational stance is that it's probably extremely rare that a fight matters. Something like "not even once a season on average per team".

This is why I'm extremely suspicious of claims of how fights give the team energy. That to me sounds like something that should matter relatively often and thus easily show up in statistics. Plus there's the problem that I also can't see it with my own eyes.

Heh, the more I think of it, the more this feels like an atheist stance. I can't see it and there's no evidence of it's existence, but somehow I'm still in the minority

(As a caveat:

As far as I've seen, most "advanced" stats are really basic stuff, beginners level. I'm open to the possibility that I'll be eventually proven wrong on this by better studies. I'm sceptical because as I said I can't see it. But it wouldn't be the first time I'm wrong.)