Run away! Run away!
Get your pitchforks and molotov cocktails ready...it's not Derrick Rose.
Every NBA blog has an MVP post. This is mine. What I'm doing to select MVP is first blending ezPM and 1-yr RAPM (these are ridge-regressed adjusted +/- data that are calculated by Jeremias Engelmann). To do the blend, I regressed ezPM onto 1-yr RAPM weighted by possessions. Here are the results:
Call:
lm(formula = EZPM ~ RAPM, data = ezpm_rapm_2011, weights = POSS)
Residuals:
Min 1Q Median 3Q Max
-438.82 -85.59 -8.76 62.90 475.10
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) -0.19937 0.12069 -1.652 0.0993 .
RAPM 0.90195 0.06381 14.135 <2e-16 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 120.4 on 402 degrees of freedom
Multiple R-squared: 0.332, Adjusted R-squared: 0.3303
F-statistic: 199.8 on 1 and 402 DF, p-value: < 2.2e-16
Notice that the R^2 value is 0.33, which is actually quite high in my opinion. You may recall that when I regressed several years of ezPM and RAPM, the R^2 was around 0.5. At the same time, it shows that a box score and +/- approach used separately may give very different results. In fact, let me show you the top 5 candidates for each metric taken alone:
ezpm
(WARP = Wins above replacement player)
| RANK | NAME | TEAM | WARP |
| 1 | LeBron James | MIA | 20.6 |
| 2 | Dwight Howard | ORL | 20.3 |
| 3 | Dwyane Wade | MIA | 16.8 |
| 4 | Kevin Love | MIN | 15.9 |
| 5 | Chris Paul | NOH | 15.5 |
1-yr RAPM
| RANK | NAME | TEAM | WARP |
| 1 | Manu Ginobili | SAS | 13.5 |
| 2 | Paul Pierce | BOS | 13.1 |
| 3 | Dirk Nowitzki | DAL | 13.0 |
| 4 | LeBron James | MIA | 12.8 |
| 5 | Chris Bosh | MIA | 12.5 |
You can see that aside from James, the lists are quite different. Ok, so back to the blending. I used the results of the regression to adjust RAPM so that it is on the same scale as ezPM. They were already pretty close to start, as you can see by the regression coefficient (0.9). I then weighted each equally and calculated WARP. Here are the top 25:
2011 Blended EZPM+RAPM
| RANK | NAME | TEAM | POS | WARP | EZPM | RAPM | COMP |
| 1 | LeBron James | MIA | 3.5 | 16.2 | 8.4 | 3.9 | 5.8 |
| 2 | Dwight Howard | ORL | 5 | 15.8 | 9.1 | 4.1 | 6.3 |
| 3 | Dwyane Wade | MIA | 2 | 13.6 | 7.0 | 3.5 | 5.0 |
| 4 | Chris Paul | NOH | 1 | 13.4 | 6.3 | 4.3 | 5.0 |
| 5 | Manu Ginobili | SAS | 2.5 | 12.8 | 5.7 | 5.8 | 5.4 |
| 6 | Paul Pierce | BOS | 3 | 12.6 | 5.0 | 4.9 | 4.6 |
| 7 | Pau Gasol | LAL | 4.5 | 12.1 | 4.6 | 3.5 | 3.8 |
| 8 | LaMarcus Aldridge | POR | 4 | 10.8 | 2.9 | 3.2 | 2.8 |
| 9 | Steve Nash | PHX | 1 | 10.7 | 4.8 | 3.6 | 3.9 |
| 10 | Kobe Bryant | LAL | 2 | 10.6 | 5.7 | 1.7 | 3.5 |
| 11 | Luol Deng | CHI | 3 | 10.6 | 2.2 | 3.7 | 2.6 |
| 12 | Kevin Love | MIN | 4.5 | 10.6 | 6.3 | -0.1 | 3.0 |
| 13 | Chris Bosh | MIA | 4.5 | 10.4 | 2.5 | 4.4 | 3.1 |
| 14 | Zach Randolph | MEM | 4 | 10.3 | 4.7 | 2.2 | 3.2 |
| 15 | Kevin Durant | OKC | 3 | 10.3 | 2.8 | 2.6 | 2.5 |
| 16 | Derrick Rose | CHI | 1 | 10.0 | 2.9 | 2.8 | 2.6 |
| 17 | Dirk Nowitzki | DAL | 4.02 | 9.7 | 1.9 | 5.8 | 3.5 |
| 18 | Blake Griffin | LAC | 4 | 9.5 | 2.9 | 1.2 | 1.9 |
| 19 | Andre Iguodala | PHI | 2.5 | 9.4 | 5.3 | 1.4 | 3.2 |
| 20 | Kyle Lowry | HOU | 1.02 | 9.2 | 2.0 | 3.4 | 2.4 |
| 21 | Nene Hilario | DEN | 4.51 | 9.1 | 3.2 | 4.0 | 3.3 |
| 22 | Lamar Odom | LAL | 4 | 8.8 | 3.3 | 2.4 | 2.6 |
| 23 | Tim Duncan | SAS | 4.5 | 8.8 | 4.1 | 3.4 | 3.5 |
| 24 | Al Horford | ATL | 4.5 | 8.7 | 3.7 | 1.2 | 2.3 |
| 25 | Andre Miller | POR | 1.02 | 8.7 | 2.9 | 2.9 | 2.7 |
I'm going to do something unorthodox and call it a tie:
2011 Co-MVPs: LeBron James & Dwight Howard
Apologies to Derrick Rose fans. He's a great player and a great narrative. But I just don't see how the numbers can justify it this season.
[...] not sure what it’s supposed to be really. My first thought was to look at the WARP (after blending ezPM and 1-yr RAPM as I did previously for the MVP race) of all players who had below average number of GS, which turned out to be 28. If [...]
[...] the other awards that I’ve done so far (MVP, DPOY, and 6th Man), I’ve blended ezPM and 1-yr RAPM, but I’m going to do it a little [...]
[...] for my previous award posts (MVP, DPOY, 6th Man, and MIP), I blended ezPM and 1-yr RAPM data and used that blend to calculated WARP [...]