RU Chop....
If we are looking at this in one game we can definitely write it off because of sample size or something that happened with the other 4 guys. When it has been 2/3 of the year and our opponents score so much less per possession when he in the game vs our players that would play over him it requires attention.
There are times when stats or metrics tell us a picture that our eyeballs don't see. Sometimes eyeballs are right and sometime eyeballs are wrong. I don't think this metric just gets dismissed.
I think your explanation, correct me if I am wrong, it just shows our D is better because Grant, Dortch and maybe Derkack isn't good on D. If that is the case I think the case is made.
The challenge with on/off is it doesn't really take into account who is coming on or off for a given player. Earlier in the year, you could have made a comparison at center, because Ogbole/Sommerville were pretty much splitting minutes. The on/off numbers would show, between those two players, how the team was doing.... but even that would be skewed, because Sommerville would mostly come on in rotations when one or both of Harper/Bailey came off, and didn't get to play alongside the two of them as much as Ogbole did.
You can (sort of) compare on/off numbers for two players at the same position, as long as they play with the same mix of players (they don't always, as with Sommerville)... but you can't rank a whole team that way. It is only a measure that shows the relative value of a player vs. the player they are replacing, not something you can compare between positions. Martini doesn't sub in for the point guard role, for instance, so comparing his on/off numbers to Harper or Davis is pretty much meaningless.
In recent games, the PF spot has been either Martini or Grant, though Martini has also had some minutes at the C spot behind Sommerville.... Martini is more defensively sound than either Sommerville/Grant, who are freshmen that have had several defensive lapses. That doesn't mean Martini's our "most impactful defensive player", though, especially since he's averaging just 15.2 min across 22 games (2 games he had 0 minutes). It just means the bigs we have behind him right now can be liabilities.
In a hypothetical situation, you could have two elite defenders who split a position 20/20, and their +/- would be pretty much identical to each other in the middle of the pack.... and at the same time, you could have another position where there was an average and a poor defender also splitting 20/20. That average defender is going to look much better in an on/off metric than the two elite defenders.
On/Off metrics have too many issues for me to take them seriously.
www.pgatour.com
