RealGM

Hey guys, as some of you may know, I'm putting together a basketball analytics blog. My first project was creating a player valuation tool which combines offensive and defensive VORP (which I've generated) to yield a final player value in terms of $. I've received some generally positive feedback, but I thought I'd run the 2012 Hawks lineup past you guys to see what you think. With that said...

Name - Offensive VORP, Defensive VORP, Total VORP - Position Adjusted Player Value
Al Horford: 566% O, 163% D, 729% T - $11.9M
Josh Smith: 439% O, 279% D, 718% T - $10.6M
Jeff Teague: 576% O, -30% D, 546% T - $7.2M
Kyle Korver: 593% O, -5% D, 598% T - $5.6M
Lou Williams: 589% O, -76% D, 513% T - $2.9M
Ivan Johnson: 379% O, 220% D, 599% T - $2.9M
Devin Harris: 468% O, -32% D, 435% T - $2.8M
Zaza Pachulia: 217% O, 51% D, 268% T - $1.2M
Mike Scott: 473% O, -99% D, 374% T - $689K
John Jenkins: 346% O, -207% D, 140% T - $492K
Johan Petro: 131% O, 201% D, 332% T - $377K
Anthony Tolliver: 69% O, -105% D, -35% T - ($86K)
DeShawn Stevenson: 33% O, -146% D, -113% T - ($293K)

This is Hawks specific, but you can find the full list here.

Thoughts? Keep in mind that player value is also driven by minutes played, so even though Lou William was great when he played, he only played 1.1K minutes (out of 3.9K possible).

My 1st thought is: if this is your creation, it's brilliant.

I've ran across VORP a few times but never fully gotten the chance to learn what it truly is. I'm sure your blog might give detail but for those that are lazy or on their phones, could you give some clarifications here? Main question first, it is "value over replacement player", correct?

It is indeed my creation! Thank you for the kind words

Yup, VORP stands for value over replacement player. In this case, I'm defining replacement player as an end-of-the-bench guy you might sign for a vet minimum, or most second round draft picks. So for instance, Josh Smith is 5.4x better offensively than a replacement level offensive guy, like Ronny Turiaf.

It's not perfect because it uses player medians, and to be included you need to have played minutes, but I think it provides fairly intuitive results (and I don't want to get into predicting how guys who haven't even made the league will produce).

This is indeed brilliant. Nicely done