TylerB wrote:qianlong wrote:TylerB wrote:RAPM is basically an unusable stat for basketball along with all other variations of +/-. It could probably be a little better for post season comparisons only but even then its probably got enormous flaws.
Could you elaborate on the subject? in particular on RAPM not the others +/- stats.
Of course it is not the gospel, and even with all the controls, I think you you have to take into account the roles played and team dynamics. Nonetheless it is a usefull stat
So you think if Collison and Bonner played the exact same role on teams that won 35 games instead of 60 they'd be in the top 20?
before i respond to some of your individual points, i'll note that i think we can agree that RAPM seems to be much better and more theoretically thought out than the more basic +/- stats. it also seems to me that it has some marginal usefulness, unlike basic +/-, though there are some pretty obvious outliers. now then...
as you mention, prior informed RAPM is SUPPOSED to be an absolute indicator of how good a player is. but it should be a red flag when matt bonner is considered by RAPM to be a top 10 player in the league offensively. the examples are numerous
Its one of those stats that for a lot of guys it is useful but for a lot of guys it doesn't work and how do you know when to cherry pick it?
i find it to be somewhat useful when you can compare how it rates the same player over several years of his career, when it's likely that the makeup of his team has undergone considerable changes. let's use LMA as an example:
https://sites.google.com/site/rapmstats ... d-march-30[/
2013 Orating 2.1 Drating 2.1 (0 is average, 5-6 is elite)
2012 N/A (only
non-prior informed data available at site provided)
2011 Orating 1.4 Drating 2.2
2010 Orating 1.7 Drating 1.8
2009 Orating 2.7 Drating 1.1
2008 Orating 0.6 Drating 0.5
2007 Orating -0.2 Drating 0.4
what stands out here? well, the 2009 Orating, for one. there is no reason to believe that he had such a big jump in offensive effectiveness when his numbers were nearly identical. in fact, his offensive numbers only jumped in 2011, yet his Orating drops to a 3 year low. makes no sense
the only thing of value i take from these numbers is that LMA has PROBABLY been pretty consistently an above-average player on both sides of the floor for his career. no more, no less
now let's look at taj gibson's career numbers, a guy who those of us having watched him his whole career would agree has been pretty damn consistent year-to-year. great defense, limited offense
2013 Orating 0.4 Drating 5.0
2012 N/A
2011 Orating -0.7 Drating 1.0
2010 Orating -1.3 Drating 0.5
alright, so pretty consistent offensive mediocrity. seems reasonable. but look at that massive jump in Drating. does that make any sense whatsoever for a guy who came into the league more or less fully developed and whose stats haven't changed much? NO IT DOESN'T. RAPM somehow totally misses the boat when it comes to taj gibson
There are big flaws with the concept. Controlling for how good your teammates are and how good the opponents are is nice in theory but it is still just an estimate
agreed. a very rough estimate. single season RAPM is highly unreliable
Joakim Noah and Larry Sanders play a similar role. Larry Sanders had a better RAPM this year. That basically makes the statistic invalid to me.
i think that's a terrible example. the only dubious part of the RAPMs on those 2 guys is noah's Drating, which is only 0.5. but sanders is widely considered one of the most impactful defenders in the game and here are their per 36 numbers:
sanders - 13/12/2 on 52% TS
noah - 12/11/4 on 53% TS
how was noah obviously the better player this year?
Brook Lopez checks in at 178th and has a negative offensive score. How do you explain that, its a flaw in the stat, not a reason to think if Brook Lopez is helping his team offensively when he is in the game. And you know its a +/- based stat some since Brooklyn has to have some guys with high scores since they were a winning team. So who does RAPM give the credit to for Brooklyn's success? Joe Johnson....who was basically horrible all year, inefficient scorer, ball-stopper, generally useless.
yeah, that's pretty ridiculous. the joe johnson part, anyway. RAPM has had brook lopez as a somewhat below average player on both sides of the ball his whole career. i'm frankly not sure that's too far off the mark. he's a below average defender by most metrics, certainly a poor rebounder and passer, but certainly a good scorer when not settling for the jumper. ballstopper? probably. but RAPM almost certainly underrates him offensively
Anthony Davis was legitimately very good last year. At times he was pretty great and if he played all season he beats Lillard for ROY I think. RAPM says he was basically the worst starter in the NBA
his defense was atrocious according to any metric that i know of. blocked some shots, but allowed guys to score left and right on him in bundles. was his offense really a -1.3? probably not. decent scorer on decent efficiency though, like lopez, not a passer so probably something of an offense stagnator
On the other hand, al-farouq aminu gets a huge score even though he struggled all year, often times was benched because he was playing with no confidence and basically was a hindrance on the floor half the time
yeah, that one looks shady
Then you just get the unexplainable stuff. Luis Scola -0.7 o-rating...1.4 d-rating....You know Luis Scola, like the worst defensive forward in the league who still gets minutes because he is such an efficient and adept offensive player?[/quote]
scola's 32 years old. he's been declining offensively for 4 consecutive years now. pretty bad efficiency. though you're right about the defense - always been terrible
i still say the main takeaway from RAPM is that it's a wildly variable single season indicator and GENERALLY respectable over a minimum of a few years. still gotta throw out for the obvious single season outliers though
