A preview of new NBA statistics

In just the past year there has been increased interest in collecting, building, and producing original statistics for the NBA that address some of the current problem areas. Mavs' owner Mark Cuban even wrote an article on specific additional data he thinks needs to be gathered and explored: Moneyball for the NBA.

Talk to people around the league and you hear many comments on the same theme:

- "I wish we had real data on individual defense"
- "assists are nice, but lots of great passes lead to free throws and the passer gets no credit"
- "I could go out there and get four rebounds a night, what I need to know is which players can get rebounds in traffic..."
- "getting players to take charges is key and something we track for our team, but I haven't seen numbers like that for the league"

The data as far as we can tell doesn't currently exist in the public realm (Harvey Pollack's brilliant stat yearbooks have some great info though). Some teams do compile a bunch of numbers, and on occasion go so far as to publicize them (the recent episode of the Pacers tracking Artest's individual defense being a case in point). By and large though it appears there are significant gaps in the available NBA statistics.

As a trial run of what we might embark on for the 2004-05 season we've been charting the conference finals at a 'kitchen sink' level of detail, hoping to understand the realities of building hard-drive consuming databases...

Here are some samples of what we are working on (all taken from the first two games of the Detroit-Indiana series):

1) Touches and play decisions
We've elected to ignore the relatively insignificant backcourt passes and focus only on the action once a team is actively on offense. With the database at hand, you can derive all sorts of interesting numbers for player tendencies and intentions.

Possession details

Team	Player	Touches	T/Min	Shoot%	Pass%	Fouled%	T/O%
DET	WallaceR	45	0.6	55.6%	26.7%	8.9%	8.9%
DET	Williamson	35	0.9	42.9%	48.6%	5.7%	2.9%
DET	Wallace	41	0.6	41.5%	46.3%	7.3%	4.9%
DET	Campbell	16	0.9	37.5%	37.5%	12.5%	12.5%
DET	Hamilton	122	1.5	28.7%	60.7%	5.7%	4.9%
DET	Okur	23	0.7	26.1%	52.2%	8.7%	13.0%
DET	Prince	39	0.6	25.6%	64.1%	5.1%	5.1%
DET	Hunter	6	0.8	16.7%	83.3%	0.0%	0.0%
DET	Billups	144	1.9	13.9%	74.3%	5.6%	6.3%
DET	James	41	1.9	9.8%	85.4%	2.4%	2.4%
IND	O'Neal	66	0.8	57.6%	30.3%	6.1%	6.1%
IND	Harrington	30	0.7	53.3%	33.3%	6.7%	6.7%
IND	Artest	94	1.1	46.8%	43.6%	4.3%	5.3%
IND	Croshere	10	0.5	40.0%	50.0%	10.0%	0.0%
IND	Bender	24	0.9	29.2%	45.8%	8.3%	16.7%
IND	Miller	78	1.2	25.6%	61.5%	7.7%	5.1%
IND	Foster	24	0.4	25.0%	75.0%	0.0%	0.0%
IND	Tinsley	112	2.2	17.0%	75.9%	2.7%	4.5%
IND	Jones	8	1.0	12.5%	87.5%	0.0%	0.0%
IND	Johnson	87	1.9	9.2%	83.9%	4.6%	2.3%

82games is hardly the originator of the notion of possession and touch stats for players, but they certainly are a powerful part of a player's makeup, especially if you believe that some of the tendencies are ingrained and hard to change.

2) Possession efficiency by play type
For some players, it's a matter of the coach calling their number on the play selection, while others make it happen on their own. By evaluating how shots, fouls drawn and turnovers came about we can get a better understanding of a player's strengths and weaknesses in scoring.

Jermaine O'Neal, Indiana Pacers

Play Type	FGM	FGA	FTM	FTA	T/O	Pts	RTG
Away from ball / intentional fouls	0	0	2	2	0	2	249
Pass into Paint / penetrate+dish / cuts	4	4	1	1	1	9	166
off Offensive Rebound	2	2	3	3	1	7	166
Drive into paint	1	3	1	1	0	3	88
Post up in the low blocks	4	14	4	4	2	12	68
Off dribble / face up J / Pick and Roll J	1	3	0	0	0	2	66
Catch & Shoot / penetrate & kick-out	1	9	0	0	0	2	22
Drive with little penetration	0	3	0	0	0	0	0
Backcourt / inbounds	0	0	0	2	0	0	0

* The RTG is a rough measure of "points per 100 possessions."

Now two games hardly makes for a sample size, but as an illustration of where we could end up going with this type of analysis the table hopefully suffices. The categories of 'Play Type' also need some refinement, but the basic idea is to measure a player's efficiency under different circumstances.

In the first two games, O'Neal was outstanding when fed the ball in the paint, or going back up with an offensive rebound, however his post game was only modestly successful, and when he elected to shoot from outside or drove the ball, got stopped but forced up a shot anyway...well the numbers speak for themselves.

Hmmm, you might be thinking 'could we isolate Jermaine's post up plays when guarded by Rasheed versus Big Ben?' Yes we can! Or how about tracking when O'Neal gets fed the ball by Tinsley specifically, or Miller? All possible!

Speaking of which, here's a look at the man who is adding wear and tear to Reggie...

Ricahrd Hamilton, Detroit Pistons

Play Type	FGM	FGA	FTM	FTA	T/O	Pts	RTG
Post up in the low blocks	0	0	2	2	0	2	249
Away from ball / intentional fouls	0	0	5	6	1	5	147
Drive into paint	4	8	3	3	0	11	122
Fast Break transitions	4	6	0	0	1	8	114
Off dribble / face up J / Pick and Roll J	4	8	0	0	0	8	100
Pass into Paint / penetrate+dish / cuts	2	2	0	0	2	4	100
Drive with little penetration	2	3	0	0	1	4	100
Catch & Shoot / penetrate & kick-out	2	7	0	0	0	4	57
Backcourt / inbounds	0	0	0	0	1	0	0

Along with picking up some cheap points at the free throw line from Reggie holding on occasion, Hamilton has been fantastic (at least by the offensive standards of this series) in driving to the hoop, in transition, and taking jumpers off the dribble. The only disappointing area in the first two games was his catch and shoot numbers being a little off.

You get the idea: we can spit out tables like this for any player -- or the team as a whole -- when we fully chart a game, and over the course of a season this kind of data would surely have some interesting tales to tell, not only about players, but about team playcalling!

We'll leave it at that for now, but part II of the "new stats" will take a look at improving passing statistics, tracking team ball movement, and adding new detail to our project to capture individual defensive efforts...