Skip to content

Updated Estimated Impact, Simple Production Score, and xRAPM

November 28, 2012

Estimated Impact and Simple Production Score

I’m gonna keep this one as short and sweet as possible and maybe expand on it another time. I was happy with the framework I came up with in attempting to evaluate the impact of individual NBA players, but not the subjective nature of the coefficients I assigned to the different variables. So I used Jeremias Engelmann’s 12 year average RAPM dataset and ran a linear regression to determine how much weight each variable should be given. Yes, something similar has been done before, but again (and by the way, you should check out Daniel M’s work, it’s really great), I think the uniqueness of Estimated Impact lies within the framework. I used all players who played at least 2,500 minutes in the 12 year period for the regression. There were 612 players in this set.

Rather than get into the complexities, I’ll give you Simple Production Score, which will serve as a snapshot of Estimated Impact best used for one game samples. SPS will give you an idea of the importance I found for box score weights and explains 90% of EI.

SPS = ((Points + 0.3 * Rebounds + 0.4 * Assists + 2.25 * Steals + 0.9 * Blocks – 0.95 * Turnovers – 0.5 * Fouls – 0.95 * FGA – 0.25 * FTA) / MP * 48) – 7


I feel obligated to point all fans of NBA player evaluation tools to Jeremias Engelmann’s newest development, xRAPM. He discusses it here, and the results (including up to date data) can be found here. Basically, it’s a hybrid of box score statistical plus minus (similar to what I have done with EI) and prior-informed RAPM, but the instead of last year’s RAPM as a prior, he has used a combination of last year’s RAPM and last year’s box score numbers. The result is a highly predictive metric. It’s a very interesting development, and I recommend checking it out.

J.E. has also calculated a fake RAPM for the 1990s (we don’t have play-by-play data for years before 2001, so he could only estimate the pre-2001 numbers with clues from quarter splits and minutes played). It’s pretty interesting and you can find that on his site as well.

Anyway, check out the Estimated Impact of players in each season up on the top right. Stay tuned.


3 Comments leave one →
  1. November 30, 2012 3:38 pm

    Heya James,
    Glad to see you back posting!

    Just a few questions.

    1. Do you have the significance numbers for your coefficients?
    2. What does regressing on 12 year RAPM give you over straight regression on wins?



    • December 12, 2012 1:29 pm

      Hey Dre, sorry it took me so long to respond, I’ve been busy with finals. Anyway, I plan on eventually doing a more detailed post so I’ll make sure to answer your questions when I get around to it.


  1. NBA Draft Projection Model and More « shut up and jam

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: