Kirk Goldsberry has opened the era of SportVU basketball analytical studies with a brilliant piece on where rebounds go, based on where the shot was taken from.
http://courtvisionanalytics.com/where-do-rebounds-go/
This is the next frontier of basketball analytics, and it looks amazing. Great job by Kirk on this piece.
Where do Rebounds Go? (Kirk Goldsberry)
Re: Where do Rebounds Go? (Kirk Goldsberry)
Hi guys,
I'm not going to pretend like I don't read this board; I do, and thanks for the nice comment, DSMok1!
SportVu data is clearly an important part of the frontier.
I think this study is interesting because it really makes us think about missed field goals in new ways. As someone pointed out in another thread, the "Kobe-assist" effect is something worth exploring.
The bottom line is that not all missed shots are created equal. Some misses are better than others; part of this can be explained by spatial approaches, part of it can't.
This study largely ignores shot context, which is a limitation.
I'm not going to pretend like I don't read this board; I do, and thanks for the nice comment, DSMok1!
SportVu data is clearly an important part of the frontier.
I think this study is interesting because it really makes us think about missed field goals in new ways. As someone pointed out in another thread, the "Kobe-assist" effect is something worth exploring.
The bottom line is that not all missed shots are created equal. Some misses are better than others; part of this can be explained by spatial approaches, part of it can't.
This study largely ignores shot context, which is a limitation.
Re: Where do Rebounds Go? (Kirk Goldsberry)
Nice to see you post here, Kirk!
What are your thoughts on the dataset so far? How great will the challenges be when attempting to use the temporal axis as well, working with the full temporal-spacial continuum for non-discrete analysis? Will using the full dataset (not just discrete events) even be useful for many things, in your opinion?
Do you have any idea if the data will be available more widely at any point?
Investigating positioning of rebounders would be very interesting--when a shot goes up, how quickly/much can the best rebounders work toward the likely rebound location, and how well can they secure it? Understanding of rebounding currently is certainly limited--who actually "gets" rebounds, contributing additional chances/possessions, and who simply receives them uncontested. Contested rebounds... there is a ton to explore with this dataset simply within this area.
I look forward to what you can do with this, Kirk!
A very interesting data source. I note you currently were using the data simply for X,Y locations of events, a relatively modest usage. The ability to spatially track everything on the court will have far more unique uses, I'm sure.SportVu data is clearly an important part of the frontier.
What are your thoughts on the dataset so far? How great will the challenges be when attempting to use the temporal axis as well, working with the full temporal-spacial continuum for non-discrete analysis? Will using the full dataset (not just discrete events) even be useful for many things, in your opinion?
Do you have any idea if the data will be available more widely at any point?
Or the Derick Rose assist, as you pointed out on Twitter.I think this study is interesting because it really makes us think about missed field goals in new ways. As someone pointed out in another thread, the "Kobe-assist" effect is something worth exploring.
The bottom line is that not all missed shots are created equal. Some misses are better than others; part of this can be explained by spatial approaches, part of it can't.
This study largely ignores shot context, which is a limitation.
Investigating positioning of rebounders would be very interesting--when a shot goes up, how quickly/much can the best rebounders work toward the likely rebound location, and how well can they secure it? Understanding of rebounding currently is certainly limited--who actually "gets" rebounds, contributing additional chances/possessions, and who simply receives them uncontested. Contested rebounds... there is a ton to explore with this dataset simply within this area.
I look forward to what you can do with this, Kirk!