That metric has a flawed integrated by using a boxscore-based prior with height, rebounding, blocked shots and steals having a huge impact. When you look through the data, you will find a heavy bias towards bigs. If the boxscore prior is replaced by a RAPM based prior, the bias is nearly gone. That points to a flawed boxscore metric, because if the boxscore metric is correct, at least the distribution shouldn't change.
We can look at the 90's for that. The calculation is based on a fake matchup file. The whole dataset without a boxscore-based prior has Michael Jordan as the #1, David Robinson as #5. Now we go through the individual years, which are based on the same fake matchup file just with the added boxscore-based prior. And all of the sudden Jordan comes not even close to Robinson, who is leading the league in all years in the 90's except for the one he was out with the injury.
Both, the boxscore-based prior and WP48 are suffering from the same problem. The coefficients were determined by examining team results. On the team level an offensive rebound is extending the possession and gets subtracted while the FGA is considered the end of a possession. That works on the team level, because players know the rule and know that not taking a shot is resulting into a shotclock violation. The team based regression does not know about that at all, and in the end basically assumes that a missed shot could have been replaced by a better shot from a teammate. That the replacement could also be a worse shot or a turnover is not considered. That ends with the funny result in both cases that a missed field goal is worse than a turnover for the individual player. Given the fact that late shot clock shots have a rather low conversation rate (well below league average) and in order to contribute as scorer in a positive fashion, he player has to score above league average efficiency, in would be beneficial for the player's individual rating to not take shot, but let the shotclock expire. That is obviously bogus, because even a bad shot has a higher probability to result into points than a turnover, thus a turnover is ALWAYS worse than a shot attempt, no matter whether the shot is missed or not.
Also, an offensive rebound is just possible, if someone missed a shot, thus, giving full credit for the offensive rebound to the rebounder is not in agreement with the specific circumstances of the game. The shooter has to get credit for not turning the ball over, no matter whether that results into a miss and a defensive rebound or a miss and an offensive rebound. Otherwise we get the insane result that scoring points of a putback is actually scoring points without using a possession on the individual player level. Which is obviously rather impossible. Berri didn't get that point in over 15 years now, unfortunately Jerry doesn't get it either, because his "new" metric ended up with a slight increase in predictive power over a non-boxscore informed version.