So Player Tracking

[EvanZ]

http://stats.nba.com/playerTracking.html

has arrived...tons of new data...touches! potential assists! Let's talk.

[mark kieffer]

I think it's cool... I am just trying to figure out what kind of information teams would value that isn't captured in the data, or isn't captured explicitly in the data.... I had a few individual research projects that are pointless to do now.... So in some ways I feel like I am at square one with determining where to go from here.

In terms of my thoughts about the data, we need more games before I can jump to any conclusions regarding how I feel about it, or what I am learning from it....

[HoopDon]

The new data is fascinating, but like Mark said, with such a limited sample, there isn't much we can do but gaze in wonder.

[steveshea]

I am excited. We need more data though. By mid year, we can run some interesting studies. I am interested in player involvement metrics. I want to know how teams (and specific lineups within the team) share the ball on offense. How does this change in "clutch time?" Does getting touches on offense give a player a little extra energy on defense? In the book Basketball Analytics (to be released very soon, see BasketballAnalyticsBook.com for more info), player involvement is measured through players' points, assists, and rebounds (since these were available). With this new SportVU data, the metric could be expanded to touches at least.

I also want to add that it's time for some more sophisticated defensive metrics. Opp FG% at rim is a nice upgrade from blocks alone.

I am interested to see others' ideas.

[AcrossTheCourt]

Oh this will be a lot of fun....

Catch and shoot percentage will be useful in comparing shoots who have to create on their own versus catch-and-shoot specialists. Guys who run offenses or create on their own a lot will have many difficult jump shots pulling down their percentages.

Pull up shots ... I wonder how much Curry will crush everyone in this category.

Rebounding opportunities are really cool. Would be interesting to compare perimeter guys who somehow manage to grab a lot of boards to interior guys.

Really cool to see hockey assists tracked.

Distance traveled per 48 minutes is a decent proxy for activity level. I should wait for a few games to say this, but some of the names and rankings are intriguing. Harden is near the bottom in distance per 48 minutes: related to his poor defense?

[EvanZ]

Pull up shots ... I wonder how much Curry will crush everyone in this category.

Not sure if you know already, but nbawowy has pullup data on the "Action" tab.

[nbo2]

Similarly to the Goldsberry "Dwight Effect" study, this is fantastic but we have to be careful how to interpret some of these ratings.

For instance, "Rim protection is defined as the defender being within five feet of the basket and within five feet of the offensive player attempting the shot," which doesn't always equate to what we think of as rim protection. If the defender is 3-5 feet behind the player taking the shot...

Then there's the issue of how often a defender gets to a spot to protect the rim, rather than forcing a low percentage when they're there. Along with the "Dwight Effect" and the "Larry Sanders Effect" from Goldsberry's study, perhaps there is an "Amir Johnson Effect" i.e. players that get to a spot to protect the rim on close shots more frequently than others. (Not too familiar with Amir Johnson's game, but from what I've heard this is one of the reasons why he is rated as a good defender by many advanced metrics).

[Crow]

Some thoughts about what could be done with this data:

FT Assists per game Secondary Assists per game are pretty low with only a few exceptions. It would be good to find the averages by position and for the league. Might want to check correlation with pace and drives.

Compute AST per game / Assist opportunities per game and averages by position and for the league. Also compute AST per game / Passes per game and Assist opportunities per game / Passes per game and averages by position and for the league.

Compute Opponent Field Goal Attempts at Rim and Opp FGP at Rim per minute averages by position and for the league. Compute difference of expected pts at rim allowed above league and position averages for each part of the equation and overall.

Compute REB Chances, Contested REB and Uncontested REB per minute (NBA continues to emphasize per game and season.)

Find player Drive / 3P FGA ratios and averages by position and for the league and for team perimeter (PG – SF) and interior (PF – C) as whole. To help characterize offensive strategy and direct defense.

For virtually every category of pass and shot stat above [insert stat] / touches to find player rates. Check especially the touch / pass ratio against the previous Chaikin estimate.

It would also be worthwhile to have Def. And Off REB Chances and Off REB Chances after 1 previous offensive rebound on that same possession.

…and drives, shots at the rim, assists and maybe more after 1 previous offensive rebound.

And summary shooting stats for drive, pull-up, catch n shoot and FT for team and PG, wing and big man subgroups.

And the assist and rebounding stats for team and PG, wing and big subgroups.


If anyone wants to take on and make publicly available any of these suggestions, or others, that would demonstrate some ability and initiative and possibly be in view of any league eyes that check around here. Of course it would be better to have 1/4th. ½ or a full season of data but an early look now or soon might be interesting if any are especially curious. I might do some of this later on, time allowing.

[Crow]

Very early (rounded) league-wide estimates using Sportsvu data


Offensive Possessions- 2.0 per minute played
Touches 1.7 per minute

Touches per Offensive Possessions- 0.85

Pass- 1.2 per minute
Pass per possession 0.6
Pass per touch 0.7

Non-Pass possessions 0.8 minute
Non-Pass touches 0.5 per minute

Close touches 0.06 per minute

Assist opportunities- 0.2 per minute
Assist opportunities per possession 0.10
Assist opportunities per touch 0.12
Assist opportunities per pass 0.17

Assist- 0.1 per minute
Assist per possession 0.05
Assist per touch 0.06
Assist per pass 0.08

FT Assists 0.01 per minute
Secondary Assists 0.02 per minute



Brief on rebounds:

For top 100 rebounders, only 20% of rebound opportunities result in contested captured rebounds. Only 1 in 3 of their total captured rebounds were contested.


Another unrelated tidbit:

Brief on team scoring balance:

4 teams have 7 guys scoring 9+ pts / game. But their team win% so far is only about 57%.

[Crow]

John Schuhmann or anyone else with nba.com:

The player tracking custom filter options could use expanding. It would help to be able to filter by team, position and more criteria.