Basketball Prospectus: Unfiltered Everything Else is Fluff.

March 25, 2009


Filed under: Uncategorized — Kevin Pelton @ 2:09 pm

In the continuing effort by stat geeks to create as many acronyms as possible, be prepared to add VA (Value Added) and EWA (Estimated Wins Added) to your vocabulary. I kid, but in truth John Hollinger‘s additions to his PER system are important and useful, taking into account playing time as well as performance in a similar fashion to my own WARP system. Check out Hollinger’s explanation of his new methods on

I thought it would be interesting to take this opportunity to see how WARP and EWA compare, at least at the top of the league. With most teams having played 70 games, the numbers have settled down and any differences reflect what the two systems value differently. Here are our top 10 leaderboards:

Player          WARP     Player           EWA

LeBron James    23.6     LeBron James    28.1
Chris Paul      21.6     Dwyane Wade     26.5
Dwyane Wade     21.5     Chris Paul      23.4
Dwight Howard   18.5     Dwight Howard   18.6
Tim Duncan      14.1     Kobe Bryant     18.2
Pau Gasol       12.7     Brandon Roy     17.0
Kobe Bryant     12.3     Tim Duncan      15.4
Brandon Roy     12.3     Dirk Nowitzki   14.4
Rajon Rondo     12.0     Pau Gasol       14.2
Yao Ming        11.2     Yao Ming        13.7

For the most part, WARP and EWA are in agreement about the league’s elite. Nine of the top 10 players are the same between the two lists, and three players show up in the same spot either way. WARP tends to favor Chris Paul‘s efficiency and Tim Duncan‘s defense, while EWA is slightly more enamored of Dwyane Wade and Kobe Bryant because of their ability to create shots.

The biggest discrepancy here is on Rajon Rondo, who comes out as the Celtics’ most valuable player by WARP but is a distant 27th in EWA. Rondo’s defense is valued more highly by my methods.

Also worth noting: EWA figures tend to be somewhat higher overall. I set replacement level using the standard that a team of replacement players would win about 10 games, which results in practice of a replacement level about 83 percent of league average. Hollinger’s empirical method, which uses different replacement levels for each position, comes out around 75 percent of league average. The standard from other sports–with the caveat that I’ve yet to get to Baseball Prospectus 2009, which tweaks the way our baseball brethern treat replacement level–is around 80 percent, so both are in that ballpark.

All of this is interesting, but worth keeping in context. I’m also working on a column that explores the debatable importance of single-number rating systems as part of a look at the big picture of APBRmetrics. Also, several people have e-mailed to ask about leaderboards for WARP and other stats from the Basketball Prospectus stats pages. Rest assured they are in our plans, though they likely will not be available this season.

No Comments

No comments yet.

RSS feed for comments on this post.

Sorry, the comment form is closed at this time.

Powered by WordPress