West's touchdown pass to Green didn't shock me . It was just...
What Does a Draft Pick Get You? – Revisited
Way, way back in the summer of 2010, we published a rather ambitious analysis on all of the post-merger draft classes, encompassing 1997 through 2009 (NBA seasons 1978 through 2010). As part of the build up to the 2012 NBA Draft (Thursday, June 28, 7:00 PM), we’re going to dust this off.
We’ll start by linking back to the original series:
- Part I: Stat Rankings and Number Crunching – The kickoff to the series gives the methodology and walks through the Top 60 picks for 33 years’ worth of draft classes. Here’s an excerpt that explains the measure being used – Production Rating (PR).
In order to do a statistical analysis on this many players, I needed something that approached a unifying number or metric. For this analysis, I dusted off an old metric developed by Martin Manley in the late 1980s called “Production Rating.” It is calculated as follows:
Production Rating (PR) = (Points + Rebounds + Assists + Blocks + Steals – Turnovers – Missed Field Goals – Missed Free Throws)/Games Played
To update this metric a little I’ve made two adjustments to it:
- Pace – I have basically adjusted all of the PR’s to a per 100 basis. As a shortcut, I used the Pace Factor for the player’s team for this adjustment. For example, Danny Granger’s 2010 numbers were “played at” 97.1, so they were multiplied by (100/97.1). It’s not perfect, but it’s sufficient for this purpose.
- Reliability – Essentially, this is just a way to adjust for games missed. The net effect is to treat the games missed as a zero (0) PR. For career reliability, I put a minimum number of years at five (5) years. This clearly doesn’t impact players whose draft classes haven’t been in the league long enough, but it is meant to penalize players who played shorter than average careers. An example would former Pacer Kenny Williams, who only played for four years. His 260 games would be divided by 410, instead of 328. If a player played five or more years, he was not penalized for “missed” years. For example, neither Michael Jordan nor David Robinson were penalized for the full seasons that they missed either at the beginning and middle of their careers, respectively.
There are flaws in this system. It will overrate stat stuffers like Shawn Marion or Troy Murphy. It will underrate players like Scottie Pippen, Joe Dumars and Shane Battier, but, hey, so does PER. Overall, however, I don’t believe that this analysis is telling you (or me) any lies.
If you want to see how others have done this, Tom Haberstroh used EWA in the D.R.A.F.T. Initiative on ESPN, and Roland Beech of 82games.com used a very simple rating combining Points, Rebounds, and Assists per game.
- Part II: Awards and Accolades – Where do MVPs get drafted? All Stars?
- Part III: First Year Impact – What kind of production does a rookie draft pick give you?
- Part IV: Validating AdjPR100 and the 5-Star System – A quick check on the relevance of the rating system.
Parts V, VI and VII were to be ranking the 33 draft classes, from “worst” to first, using a the following point system based on the 5-Star ratings and the Peak Award levels:
Parts V & VI were completed, and the links are below.
Part VII will be coming out (better late than never) by tomorrow, in order to provide a complete series.
From there, we’ll be updating the information for the last two years, and doing some analysis more relevant to the upcoming draft.
- What kind of production has come from the #26 pick, as well as those taken after that slot? We’ll be investigating whether later picks have become more productive over the last few years. Some anecdotal evidence exists, as several of the most productive #26 picks have been recent ones – including Pacer and Indianapolis native, George Hill
- How have Pacer execs Larry Bird and Kevin Pritchard done in the draft. For Bird, we’ll examine all of the Pacer picks since 2004 – as well as a discussion of draft day trades. For Pritchard, we’ll look at his tenure as the GM of the Portland Trailblazers, spanning from 2007 through 2009.
- Updating the analysis to include the last two years. What do the classes of 2010 & 2011 look like? Have the last two years of data made any significant changes? It will also be worth spending some time taking a look at the Class of 2008 – which has generated enough success over the last two seasons to have placed it 6th among all the (now) 35 classes based on the system above.
These will come out over the next 10 days or so leading up to the draft.