Can RAPM and plus minus metrics be "inflated"?

[Heej]

The discussion in that thread is silly. This idea that Lebron's teams are more interested in making Lebron's stats look as impressive as possible over trying to win games and championships is absolutely ludicrous.

But if one was so pro-Mike they were willing to talk themselves into such a narrative, they would have to look at how incredibly successful Lebron's teams have been despite the teams not actually prioritizing winning, that it actually would strengthen rather than weaken his case against Mike.

The bottom line is Lebron James is one hell of a basketball player and that's why his impact stats look great. Because he is having great impact. And its not zero sum so his impact has nothing to do with Mike.

I would just completely ignore that whole line of argument.

The funny thing is during LeBron's Heat days the Heat scorekeepers were known for being among the stingiest in the league when it came to assists. If they were generous like the Clippers' iirc at the time were known to be he likely would've set the PER record

[rcontador]

RAPM seriously underrates clutch play.

For instance if "high motor" Player A adds four points to his team every single game, while "clutch" Player B adds ten points to his team in end-of-game situations in every close game, they probably net out to roughly the same RAPM. But Player B is 2.5 times more valuable in terms of winning, i.e. he's a much better player.

I assume this bias often cancels itself out when comparing real players, but it's one of the central theoretical limitations of RAPM. In particular, it's a major reason to discount defensive RAPM compared to offensive RAPM, since offensive skill translates better to clutch situations.

A player who is +10 in close games will also be +10 or close to it in less close games, and if he somehow is a zero in all games except close ones (?) Then his value is not really thst of a +10 player on average

That hipothetical is not realistic

Yeah, it was a thought experiment.

I literally said that in the real world a lot of the error washes out. (But, importantly, not when comparing offensive versus defensive players!)