If you start with your conclusion, you end up with a lot of bad ideas. If you start with an open mind and use the data to understand what happened you get a better picture.
I don't think that's fair to say. Quite the opposite, I think you might be arriving at your conclusion based on a stat that may not be accurate. Hence, the Matt Donovan dilemma. Great stats but he's just struggling so much. Possession and CORSI aside, he carries the puck well, makes smart passes, but his ability to cough up the puck in an instance, or be slow to respond to an on-ice situation (break-out or transition) or blown coverage, leads to a lot of scoring chances and goals against - specifically HIS FAULT.
That's eye test BUT there's supporting video evidence of the SPECIFIC mistakes. That's FAR MORE reliable than general stats - NO?
Yes, you can use the stats, but in the right way.
Chicago Blackhawks aren't built that way. I simply do not accept that "the way things have been done by most" is necessarily the best way to do things. Thinking outside the box sometimes leads to better outcomes and the only way to find out is to test it.
That's a bit of a blanket statement though. Sumo wrestling goalies are also very "out of the box". To suggest something is different doesn't make it better, necessarily.
I'm in favour of advanced stats, absolutely. My skepticism is based on the stats themselves, as they are.
Essentially, these stats (CORSI, Fenwick) aren't good enough. Nowhere close to eye test and video review for specific coaching and analysis.
What would you rely on, if the game is so complex? If the answer is "my eyes" then I would point out that trying to keep track of everything without writing it down in a complex situation is very challenging and highly unreliable.
Eyes and video the vast majority of the time.
Stats can be used in a small way to gain insights, possibly. But the current stats are highly unreliable and unproven as far as I can see.
Well, here's one way: Jonathan Towes relative CF% is +5.75 for a 59.4% CF in even strength. Blake Wheeler is a relative CF% of -0.61% for a 49.9% CF% in even strength.
Toews is an all-world defensive player and it shows in his CF rel. Wheeler is not, and it also shows in his CF rel.
It's unfair to compare CORSI REL from a good team to a bad team, no?
So I've got some more relevant examples below
I'm not going to speak in platitudes here, but the grain of salt here is a hell of a lot smaller than in any other analysis; especially the promise of what your eyes tell you.
You basically said above that there is no stat to tell us why Toews is so much better than Wheeler. I pointed to a stat, the stat I use all the time in similar discussions. If anything, that should cause you to rethink what you've been saying here. There is no excuse not to.
here's a platitude:
Kane's Rel Corsi is the 10th lowest among forwards in CHI. The leader is Jeremy Morin. Ahead of Kane are guys like Peter Regin and Andrew Shaw.
how reliable is this stat? or maybe Kane should be relegated to the 4th line and only play PP? He's not in the line-up to kill penalties or play physical.
There are such examples on all teams. One the NYR, Brad Richards had a far higher CORSI REL than Stepan and St. Louis - maybe they should have kept Richards around?
TEN forwards on the Islanders had a higher CORSI REL than John Tavares, on track for over 90pts before he got hurt.
Is it possible that CORSI Rel is NOT a reliable statistic? More precisely, it's possible that possession time or pucks directed at net/on net while player "X" is on vs. when that player is not on really doesn't tell you much about how effective that player is on the ice? By effective, I mean in generating net offense (offensive chances, or goals/assists or points).
Maybe the correlation is NOT related to causation?
Like the full moon doesn't really cause people to commit more crime?
I don't mean to be facetious. I just don't see the straight correlation between CORSI and player effectiveness.
I would like to hear your response to the Kane example (and there are many many others)
While it's one thing to suggest "well, the stats are accurate so let's see what Jeremy Morin is doing right and Kane is doing wrong" but that seems like a not-so-smart way to interpret the stats, no?