APBRmetrics

[fundamentallysound]

I went to go check out some RAPM numbers for a post I was going to write on BlogaBull and it appears HoopNumbers is totally down. Anyone know what happened to it?

Is jsill still hanging out here?

[Ilardi]

I wondered the same thing a few weeks ago, as it appears the site is down (understandably, since Joe is now consulting for an NBA team).

However, I'm guessing most of the interesting content is easily recoverable via Google's cache function or other web archives like 'wayback'. For example, in less than a minute of searching I was able to find these Hoopnumbers team-by-team 09-10 'regularized' APM numbers: http://webcache.googleusercontent.com/search?q=cache:Tj2oU_HnzQ8J:www.hoopnumbers.com/allAnalysisView%3Fanalysis%3DRAPM+site:www.hoopnumbers.com+hoopnumbers&cd=3&hl=en&ct=clnk&gl=us

[Serhat Ugur (hoopseng)]

[Ilardi]

I don't know if it's confidential or not, so I'd rather err on the side of caution.

[Ilardi]

[DSMok1]

It's too bad he didn't leave it out there... It was public, is it no longer? It was rather useful.

[Ilardi]

When I did some ad hoc consulting with an NBA team for a few months back in 2008-09, they asked me not to publish anything new, but they certainly never asked me to take down work that I'd already put out there in the public domain. It wouldn't really surprise me, though.

[jim]

[Ilardi]

No, I'm not working for anyone in the NBA right now. I've just been too busy in my "day job" of late to get much done on the hoops front . . .

[schtevie]

I look forward to finding out what team Joe is consulting for because there are bucket loads of bad karma heading its way, if in fact there was an official request to cleanse the web of historical data.

It is not as though the method generating the data was proprietary in any meaningful sense, and the specific estimates themselves are of rapidly fading value.

This is really bad form and a thousand curses upon this franchise to be named later - shame on it (unless it was the Celtics, in which case I retract the comment)!

[Ilardi]

I certainly understand the sentiment, Schtevie, but I can also appreciate that a team may feel that Joe's nuanced approach to ridge-based APM gives them a competitive advantage worth protecting . . . There are probably only a few dozen people out there in the NBA analytics community who can recapitulate Joe's analysis in the absence of his published work on the topic, and apparently there are *no* current APM stats out there in the public domain other than Aaron's 2-year numbers (which are pretty noisy).

[schtevie]

Steve, you are ever gracious and too kind. I think you are bending too far backwards to come up with a reasonable interpretation for the scrubbing of the available data.

RAPM indeed offers interesting and intriguing results, but it is by no stretch of the imagination the final (or even currently ultimate) word on the estimation of player value. Furthermore, any team intrigued by the approach can with relative ease and low cost find someone with the technical skills to provide up to date results. And the latest estimates provided on hoopnumbers, if I recall correctly were partial numbers for 2009-10!

So, who is hurt by its elimination? Statistically/historically-inclined fans; future Joe Sills; and the collective community - the specific NBA franchise included - who might find some benefit from the sharing of the previously non-proprietary data.

Now, I don't know if anyone told Joe to take his ball and go home. This is speculation. But it is really shabby, small-minded behavior, and there is no getting around this point.

By contrast, one should not forget to praise those franchises and individuals who have chosen not to suppress once public knowledge (such as the Houston Rockets, which let Eli W's invaluable countthebasket posts stand) but also have allowed the continued provision of public knowledge. Hat's off to the Mavs and Roland for 82games (and was it Golden State before?), the Grizzlies and Aaron for basketballvalue, the Pacers and Kevin P for basketballprospectus, and Portland and Justin for basketball-reference (with apologies for any omissions or inaccuracies on this list).

Imagine the poverty of our existence if all these sources were to vanish.

P.S. I should add that I hope that there is a benign explanation for the apparent action taken, and hopefully it can be reversed. Otherwise, this as precedent augurs a more dismal future.

[jsill]

Hi, guys. For the record:

1) The team I'm consulting for did not tell me to do anything regarding my website. So any "shabby, small-minded behavior" is entirely my own and not my employer's (maybe I should add that as an automatic disclaimer signature at the bottom of all my posts).

2) I did indeed take the site down. I'll consider resurrecting it at some point, although I don't know for sure what I'll do.

3) I definitely did not ask Google or anyone else who archives the web to scrub their archives of hoopnumbers pages. You'll notice, for instance, that the Google cache link which Steve Ilardi posted earlier in this thread which lists the single-year team-by-team results still works. Also, here's another page from my site (the writeup of my four-factors approach) which is still available via Google cache:

http://webcache.googleusercontent.com/search?q=cache:ryDlDEafhRkJ:www.nbastuffer.com/index.php%3Foption%3Dcom_bookmarks%26Itemid%3D168%26task%3Dview%26id%3D107+hoopnumbers.com&cd=4&hl=en&ct=clnk&gl=us

I can't say for sure why it became harder to find the pages from hoopnumbers through the Google cache, but my guess would be that Google's search algorithms stop listing a particular site in the results if a site has been down long enough. This would certainly be a reasonable approach, since Google doesn't want to be producing search results which link to dead sites.

I'm sorry I inconvenienced folks by taking the site down. If anyone who isn't with another NBA team wants to check some results against mine, I should be able to send you some of my results via email or otherwise give you a few tips.

[Ilardi]

Joe,

Thanks for weighing in.

As I mentioned above, it's completely understandable for a team (or a stat consultant) to wish to keep a competitive advantage, so I certainly don't regard the decision to take your site down as indicative of either "shabby" or "small-minded" behavior. It's your prerogative, after all . . .

Interestingly, the day after I posted the google cache link to your site (above) that same link stopped working. But I see it's working again at the moment, so I can only guess that there's some flux in the accessible content of google's cache for a few weeks once a site comes down.

In any event, thanks for the clarification, and best of luck in the future.

Steve

[schtevie]

I would like to apologize to Joe for the excesses of my rhetoric. "Shabby" and "small-minded" are words that were too strong and imprecisely employed, though the removal of publicly available, quality research is something that I strongly oppose in principle and in the instance.

And best of luck to him in his consulting endeavors (especially if it is for the Celtics)