... sorry about my inability to edit...
For a lot of things, you don't have to be the best to have success, you just have to be good enough. Michael Jordan wasn't the tallest guy in the NBA, but he was tall enough to make it. He wouldn't have been much better if you just made him 1 inch taller, for example.
But see, if we're talking in the 3-5% range of differences here, 1 inch on Jordan is a lot less than that. Taking 5% of Jordan's height from him makes him closer to 6'3", and all of a sudden we're starting to rule out having the height and wingspan to dunk from the free throw line, for example... But Jordan may have made way more than 5% more of those "impossible", sometimes game-winning shots than you'd expect the next best guy to (which could be the difference between 4 wins/8 points over an 80 game schedule), so you've got a real apples and oranges attempted comparison here.
Point taken, but I think it loses its bite by attempting to lump the "best" in with "average".
And look at the data, there's absolutely no correlation between overall team FO% and wins.
If there's supposedly a really strong correlation to possession and winning (hence all the CORSI champions), then all that remains to be done is to link FO% to possession, right? I don't foresee a huge leap in logic being necessary, but I do foresee the ability to track possession between faceoffs in the near future, so... TBD.
As I said, it's nice to have a 60-65% guy for specific missions, but otherwise, the puck is turned over or dumped so often during a shift that winning the FO is pretty much useless, especially in the neutral zone.
Then, from a coaching aspect, you "simply" address the turnover aspect. Much better starting point than devising forechecking schemes tailored to every opponent in order to constantly
regain possession. The ability to choose where the puck goes obviously can't be overstated, as even dumping the puck in after a neutral zone draw keeps at least a couple of their players on the ice while allowing you the option to substitute any number of yours on the fly.
That's why all these advances stats are pretty and interesting in their novelty, but still fail to encapsulate/measure the "effectiveness" of any choices that are made during a pretty fluid game like hockey. Sometimes one relinquishing of possession allows the chance to make one change that affects how chances might be generated for the next 2 or 3 shifts. It doesn't take a certain percentage of success:countable events. It takes, on average, the ability to generate 3 or 4 such chances on any given night (more technically, at least one more than your opponent), and it doesn't matter which choices ultimately lead you there. The stats certainly won't remember that aspect either (choices/strategy set between goals), which is where those with the trained eye are relied for the "test".