Sorry, an error occurred while loading the content.

## Re: [APBR_analysis] Masters Thesis

Expand Messages
• ... Well, you are in luck. Most of the work in NBA stats have used traditional box score numbers, but for the last few seasons, the NBA has published game
Message 1 of 2 , Sep 27, 2004
ivan ivan wrote:
> I'm a Math Graduate Student at SFSU
> And I want to do my thesis on the NBA
> I've been looking into Patterns of winning teams
> and I 'm doing a logistic model on the basic stats to see how
> valuable each stat is compared to another i.e. relative value of an
> offensive rebound compared to a steal
> but I need more
> I need to do some deeper analysis
> or at least some more original work
> so that's where you guys come in
> You seem to be doing alot of analysis on the NBA
> and I was wondering if you had any project you might be working on
> that you wanted some help with
> I'm very good with data analysis
> and programing in SAS
> I just need some help with direction
>

Well, you are in luck. Most of the work in NBA stats have used traditional
box score numbers, but for the last few seasons, the NBA has published game
play-by-play logs. These have been used by a few people to generate new ways
of looking at what happens on the court (82games. com,
basketballmetrics.com, not to mention DanR at
http://www.uncg.edu/bae/people/rosenbaum/NBA/winval2.htm) but this is just
scratching the surface of what these logs can be used for. Play by play
analysis is really in its infancy -- there's a lot of fertile ground here.

Personally, for some time I have been trying to assemble a complex game
state matrix so I can do analysis similar to Jay Bennett's Player Game
Percentages used for baseball (http://www.amstat.org/sections/sis/pgp/).
It's pretty slow going for me as I lack the programming skills. If you want
more information as to how this type of analysis can work, or other
directions to which it can be taken, contact me off list.

ed
Your message has been successfully submitted and would be delivered to recipients shortly.