RAPM

[Intza]

I have a question about this stat. On the ESPN page I have been watching the defensive numbers.

I have noticed that the best defensive point guard score around +2 and the worst around -5. For centers, the best are around +5 and the worst around -2.

¿What do you take out of this?

Since the difference for both is 7 points, ¿does this mean that the diference between a good and a bad defensive player for centers and points guards is basically the same?

Popular knowedge tells us that a good big man has more impact than a good perimeter defender, and the stat supports that.

But also tells us that a bad interior defender kills your defense, and this numbers say that a bad point guard defender is even worse.

I don't understand the math behind it, so maybe I'm dead wrong, ¿what do you think?

[mysticbb]

I don't understand the math behind it, so maybe I'm dead wrong, ¿what do you think?

I think you are quite right. Not that it should differ a lot, but when we look at RAPM (which RPM is essentially just a derivate of) in a pure version shows that the difference between great defender and bad defender on each position is rather the way like you described it (bigger for bigs, smaller for smalls). For RPM that get changed by using a prior which is partly boxscore-based. In that way defensive value gets shifted to the defensive rebounder and to bigger player, thus resulting in the "equalizing of difference for each position" like you observed. Though, the used boxscore-based metric does improve in average the predictive power, but it comes at such a price.