Friday, May 21, 2010
John Hollinger's computer is talking smack
By Henry Abbott
Well, not really. I'm pretty sure his computer doesn't talk at all. And Hollinger is too thoughtful to really brag.
But do you see what's going on here? For all the moaning about how basketball scouting is far too complicated for spreadsheets, for all the boilerplate "Moneyball sucks" conviction, for all the appreciation there may be in this world for things like heart and fit and dedication to craft ... the fact remains that past performance is the best indicator of future results.
And in judging past performance, Hollinger's spreadsheet-based system is handy. How handy? Well, if you fired your entire scouting staff and front office and just drafted based on Hollinger's ratings, you'd probably be better off. Hollinger does not advocate that, by the way, and he acknowledges his own tool could be a lot better.
But the memo is: Don't pat yourself on the back too hard for ignoring smart objective evidence like this.
Here Hollinger describes his draft rater (Insider), and how it has outperformed actual NBA front offices in recent years.
It's based on regression analysis that compares college performance using 27 different variables, from the obvious (age, height, likely pro position) to the obscure (say, 3-point attempts per field goal attempt). By looking at what talents have led to success at the pro level, we can figure out some things about what current collegiate draft prospects will be able to do in the NBA.
My original motivation was the fact that the actual NBA draft has produced so many busts in the top 10. While the Draft Rater has also uncorked a few (you can see the history through 2008 here and last year's list here), the overall results have have been strong.
The Draft Rater has yet to miss a lottery pick who became an All-Star in its top 12 collegians list -- although that string may end in another year or two thanks to a miserable 2008 performance (Russell Westbrook and Brook Lopez both were overlooked that year). And if it's blown a couple of picks, look at the actual draft and you'll find even more mistakes by the pro teams themselves.
On the other hand, the Draft Rater has picked out five All-Stars that the pros missed among the first 12 collegians -- Carlos Boozer, Rajon Rondo, Danny Granger, Josh Howard, and David West. No misses, five additions. I like that ratio.
Last year his system loved Ty Lawson, who went on to be one of the best value picks in the draft. This year, DeMarcus Cousins is the star of the class.
One day it would be nice to have sophisticated rankings of different front offices and their draft performance (tricky, because you never know if it's the owner or the GM really making the call, but still ...), then to see how Hollinger's laptop stacks up. We may yet get that thing to talk a little trash.