Play Analysis: import of player / lineup / other components

[Crow]

Question guy here. What about analyzing play types and their success at team or league level by the components present in them at player or lineup levels or details thereof? Anybody done that, want to discuss or do that? Is pick n roll / ballhandler most influenced by a,b,c lmnop, xyz, 123, etc.? Is spot up shot more influenced by post threat or quality of other shooters or player speed / agility, experience or elevation, quality of teammate picks, lineup pace, quantity & speed of passing or ? I don"t recall seeing anything like. If play type is important or most important framework to work in (at least for coaches and compared to player value rollups) it is probably time for this.

[jgoldstein34]

So correct me if I'm understanding your question wrong, but you're pretty much asking to quantify the importance of different types of gravity on each play type? For example, if you have a great spot up shooter and a great roll man, what's the impact from each on getting a 3rd player open for a shot after a pick and roll? It sounds like a really interesting question, if I'm understanding it right.

My guess would be the impact of a shooter vs roll man is similar and what matters more is how good the individual player is at their role. i.e. DeAndre Jordan and JJ Redick are both elite at what they do and probably help draw open space for Blake Griffin in equal amounts for a mid-range shot.

What matters most is probably how easily a defender can help off a player and then recover back in time to contest a shot/layup.

[Crow]

Gravity, yeah; or "play plus minus". And yes it could /should be done for defense too.

Need the raw Synergy data for best implementation. Maybe could get part way there for simple, obvious components with a good self-made play by play database.

[Nate]

Question guy here. What about analyzing play types and their success at team or league level by the components present in them at player or lineup levels or details thereof? ....

Isn't that basically what this sloan paper is about: http://www.sloansportsconference.com/wp ... 2/1624.pdf

Doing that kind of analysis is going to be a lot of work because it's like you're asking about the weather vane on top of a building where people haven't even laid out the foundation.

We don't have answers to fundamental questions like "what are the play types?" or "what are the measureable dimensions of player quality?" so when you ask "does player agility impact the rate of corner 3 shots on off-ball pick plays?" the question doesn't even make sense yet.

Someone who starts with an a priori taxonomy of plays and notions of player quality could do a bunch of work to build and test hypotheses like that, but it's unlikely to be repeatable.

[Crow]

Yes the paper is in the same direction. It is just a beginning. It is time for more. A lot more.

[Rd11490]

Is the location data even publicly available anymore?

[Nate]

Is the location data even publicly available anymore?

I'm not sure more than this was ever published.

https://github.com/sealneaward/nba-movement-data

[Rd11490]

Thank you so much for this! I did not realize people still had copies of this floating around.