Dr Olajuwon wrote:Guys, what do you think about the OBPM, DBPM and BMP stats? How well do you think the BPM stat reflects the real impact of a player? Do you like to use it to get an idea of a player's impact in the game?
Well let's start with the obvious: it's an equation designed to match as closely as possible the results of RAPM, without any input from RAPM. Meaning, it weights all of the individual box score stats such that the result is the closest model of RAPM possible using this limited sample.
It just has all of the same flaws as using any box score metric. It's based on an arbitrary, biased sample (Chris Paul and pre-trade Rondo averaged the same number of assists; are they equal playmakers?) and does an astonishingly horrible job evaluating defense. Like, seriously, look at how that stuff is calculated. BLK%, STL%, and DRB% are included in
offensive BPM, and then DBPM is taken by BPM - OBPM.
It's probably the best we're going to do if we continue to just build things based on the box score. But that's not saying much, given that the box score has been shown time and time again to not properly assess a player's impact. We have a litany of much better and more useful impact stats now and on the horizon, and quite simply the only reason BPM would be useful is because it's the easiest to find (and I'm not discounting this, it's certainly an advantage). I won't use it in any type of serious analysis, to be frank.
EDIT: Now that I'm re-reading this, it's also kind of telling that these guys looked at the box score, and said "the best we can do with this is approximate RAPM". That's funny, but also very telling.
(just mentioning those who were recently active in the project).
