Disclaimer: Box Score +/- is in no way an end all be all stat. It punishes a guy like Pav who simply never comes off the floor and fails to account for who is on the floor for the other team.
Not an expert on box score +/- but always thought it has some flaws like the one you mentioned. Similarly, it also fails to account for who is on the floor from your own team, no?
Correct, it can be a very flawed stat, but the Kuany/Hadaway 1:1 substitution pattern combined with those numbers felt relevant. There's an up-and-coming college basketball data guru who created a site,
https://evanmiya.com , that I've always found really interesting. The numbers and player ratings he produces are great because he accounts for all of the above, while also doing things like excluding "junk time". His methodology accounts for many of the historical shortcomings of metrics like KenPom.
To further back up the Kuany vs. Hadaway point, the site also includes breakdowns for each lineup and combination of players. The easiest way to compare them is when each player is on the floor with the starting four of Pav, Simmons, Kelly, and Sheldon. The data below includes each lineup's +/- per 100 possessions, as well as their adjusted efficiency ratings per 100 possessions for both offense and defense. These adjusted efficiency ratings account for matchups, opponent strength, and they exclude junk time.
Hadaway / Pav / Simmons / Kelly / Sheldon
+7.6 per 100 possessions (not adjusted for opponent)
115.2 Offensive Efficiency
102.5 Defensive Efficiency
+12.7 Efficiency Margin (adjusted for opponent)
Kuany / Pav / Simmons / Kelly / Sheldon (Smaller sample size)
-43.75 per 100 possessions (not adjusted for opponent)
115.7 Offensive Efficiency
148.7 Defensive Efficiency
-33.0 Efficiency Margin (adjusted for opponent)