MChambers
Member
Bart Torvik agrees that some part of "luck" is just what metrics don't capture, in some cases, but he certainly doesn't go as far as you: https://adamcwisports.blogspot.com/search/label/FUNAlas I am fairly certain, without looking at the data because I'm lazy, that there isn't typically a lot of "regression to the mean" either. More specifically, there are teams who stay "lucky" or "unlucky" all year long. Kenpom's luck is perhaps best viewed as capturing what metrics cannot. As such, it may be a very useful hedge when selecting tournament winners, etc. An hypothesis begging to be tested with the right weighting of metrics and luck factored in. Pardon the musings.
As for how to measure "coaching," I do believe the DBR method of loudest opinion seems most prudent!
I'm avoiding the word "luck" here for a reason. If I call it luck, that implies that the T-Rank is a rather absolute reflection of a team's quality, and that any deviation from T-Rank's expected winning percentage is simply random variance, or luck.
Undoubtedly, that is a big part of the story. But it's probably not the whole story.
If you were right, then W/L record would be more predictive than it is.