Doctor MJ wrote:runtmc wrote:slick_watts wrote:dean oliver is an analytics legend but imo a bit of a dinosaur.
Dean is certainly a legend, one of the early members of the basketball stats community. Aside from Basketball on Paper, he started the APBR message board on the old sonicscentral forum and was an early influential member, along with Dan Rosenbaum, Kevin Pelton, Roland Beech, and to a lesser extent Hollinger (and others of course). He was the first guy from APBR to get hired as an analyst by a front office, and he also helped organize the Sloan conference. I talked stats with Dean a bunch back in the day on the APBR forum when I was still a stats undergrad and can say first hand he's genuinely a very smart guy. He studied math or physics (I dont remember) at CalTech and had a Phd in stats. Suffice to say he knows his stuff.
Back then, there were mostly two camps: the possession based stats camp was lead by Oliver, and +/- based stats was Rosenbaum's camp -- it seems like this is more work along the same vein. He hasnt done much publicly in basketball stats in years, but I dont think its fair to say hes a dinosaur -- its just most of the work he's done has been behind closed doors. He's been in the industry/front offices for a couple decades now, and Im quite sure he's still involved with a lot of the latest stuff going on in analytics.
As far as this metric, its hard to say anything about it, given there isnt anything about the methodology I can find. But given its Oliver's work, it seems like it should be useful, its just hard to know how useful/in what contexts without more information.
Interesting your description of the 2 camps. When I first joined RealGM I also soon joined APBRmetrics kind of expecting I'd get more into that community than this one, but while I had a lot of respect for Oliver and the the other big names there, it felt like most of the actual community there really was just doing Tendex/PER-type stuff and that they were actually hostile to the use of +/-. Yes Rosenbaum was a sometimes poster and he was involved with +/- and respected, but among the rank & file, it seemed like a group that was fighting the previous war.
Gradually as PM led to APM & RAPM, and the guys doing that stuff got hired by NBA teams, APBRmetrics became a major place for +/- style stats, but by that point I was busy doing other career stuff and RealGM was the place that felt like home.
In terms of the possession-based stats camp, it's interesting because I didn't perceive that as a separate camp. Oliver was the main guy associated with it without question, but I felt like he was universally appreciated for that work by both the Tendexers and PMers.
Does my recollection fit with yours and do you feel like I had something backwards?
Ah cool, there's a few of us still kicking around it seems.
Funny how our experiences color our perception/perspective of things. At the time, I was a stats undergrad and was interested in/doing work on possession based stuff in my free time, so that might be the source of my perspective on there being two camps -- Oliver was someone that I went to for feedback on the stuff I was doing, along with a few other posters (wizardskev/Kevin Pelton was another one), while Rosenbaum and the rest seemed more interested in pursuing his work on RAPM at the time (he had just come out with his RAPM stuff a few months after I joined, and a lot of people jumped on the bandwagon). At least at the time, I dont remember much work on Tendex/PER type stuff -- I remember there was a bit of talk when the wages on wins stuff came out, but that was more towards the end of when I was there, and some people did some work on stuff around that. I just dont remember much work being done with weighted metrics. Or maybe I just didnt pay much attention to it -- I didnt think much of it at the time. Also I think it sounds like you might've come along a little bit after I did? I thought it was ~2002/2003ish when I joined, but Basketball on Paper didnt come out until 2002 and Berri's stuff didnt get published until 06 apparently (although I vaguely recall he posted the wages on wins stuff to the board before he later published, so Im not sure). It must've been more like 04-06? when I was active, but the timeline of things is a bit hazy honestly, its been awhile.
One of the things I got involved with was the project for charting play by play data for Roland from 82games, but something that bothered me was they were going to keep the data private. If you didnt participate in charting, you werent going to have access to the data, and that rubbed me the wrong way -- I felt like we should be making the data public. There was a lot of trading of spreadsheets of raw data behind the scenes/in DMs -- wizardskev actually shared a bunch of data with me, or I wouldnt have had access to much of it at the time. There was a site -- I think it was bball index -- who later posted a bunch of the raw play by play data, and I did a project at school using some of that data. Anyway, between all the politics about data, Oliver leaving for a front office job (Pelton later did too), and me getting busier with finishing school, I drifted away from APBR after a couple years and started mostly just lurking occasionally. I also got contacted quite a few times in private from people who wanted me to convert some of the work I had done into prediction stuff for sports gambling -- I probably got at least a half dozen different inquiries -- and that also had turned me off. I stayed in contact with a few guys via email for a couple years, but slowly lost contact with everyone.
Do you remember what time period you were on there/did you post much? APBR was such a small community at the time (there were probably less than 50-100 regular posters, although Im sure there were a lot of lurkers) I didnt realize there was that much crossover between there/RealGM, so thats nice to see.
