Introducing the Box Model
by Ziller on Jun 9, 2009 7:38 AM PDT
in Commentary
40 comments
I think visually, and to that extent spatially. I've been thinking about this particular idea for a while -- how can we explain a team's performance in a completely visual way? It's not perfect (yet), but I think I have something here: the box model.
Everything an offense can do is in here. (Conversely, you could do an identical one for defense. I might flip it upside down, though.) Remember the Anatomy of a Possession? All of that is in here. Everything good or bad fits into offensive rating, and it all fits right here. (This is one reason I chafe at criticisms that advanced statistics do not capture everything that happens in a basketball game. Capturing is not the problem -- the final score of a game actually captures everything. The problem with advanced statistics right now is one of identification -- how do we pull out the information we need without also getting information which might cloud our judgment? But the capturing of total information, we have that.)
The red field represents turnovers. It grows right to left -- you can the Kings were very slightly worse than average in turnover rate this season. The blue field is point scored. The big deal, the bread and butter of an offense. This can grow two ways. First, a decreased in turnovers (red) opens up the field for shots (blue, green, black) more. Second, a team's shooting percentage affects the height of the field. I use True Shooting percentage because ...
... it includes adjustments for three-pointers and free throws. So those free throw possessions -- those are in here. A team which draws more fouls will see a bump in shooting percentage, which grows the blue field. Offensive rebounds are represented by the green field. There's an important distinction in its location relative to turnovers. You might think that because turnovers represent shots lost and offensive rebounds represent shots gained, the rebounds should be a vertical field on the left. But offensive rebounds only help when you miss a shot. If you make every shot for 5 minutes, your elite offensive rebounder won't help you. If you turn the ball over three straight possessions, offensive rebounding ain't going to help you. This isn't to say offensive rebounding isn't important -- it is. But it only comes into play when there's a missed shot.
What about other factors of the game?
* Assists. Those fall into shooting percentage. Though there is no discernible relationship between assists and shooting at the team level, studies have shown a decent increase in individual shooting percentage when a potential assist has been made, something like a 8-10% boost. (So someone who shoots 30% on pull-up threes would theoretically hit 40% of his catch-and-shoot threes. Theoretically.)
* Screens. Again, shooting percentage. Screens in particular get pointed at as a facet of the game stats can never capture. And no, there's no public statistical assessment of the value of screens. But they are counted in the big picture, assuming there's a substantial of a good screen versus a bad screen (a theory I subscribe to).
* Fast break. Shooting percentage, and potentially turnovers.
***
As we can see from the specific Kings offense box model, the fundamentals of the Kings offense are strong. The team shoots above league average (thanks, Kevin), the turnovers are not completely out of whack. Simply by virtue of Spencer Hawes becoming a more efficient offensive player (which I think we all believe will happen) and by virtue of the guards (including Kevin Martin -- it's not all on Beno Udrih) turning the ball over less frequently, the offense will improve. It was No. 25 in the league last year, but it can be much better, assuming some health for Martin and (again) improvement for Hawes. Replacing Mikki Moore with Jason Thompson for a full season will help. The No. 4 pick might not help that much, but it could be a big immediate boon. We'll see, but I'm fairly confident in the ability of this offense to get much, much better.
The defense, on the other hand ...
4 recs |
40 comments
|
Email
Print
Comments
It's beautiful
But what I’m curious to see is how the Kings’ Box compares to the boxes of other teams.
Never forget: I am a complete idiot
by Exhibit G on Jun 9, 2009 7:58 AM PDT reply actions 0 recs
I'm curious to see the the Kings Dance Team's box
www.mancancook.net
by vfettke on Jun 9, 2009 8:16 AM PDT up reply actions 0 recs
I just want to look at boxes
Is that too much to ask?
Never forget: I am a complete idiot
by Exhibit G on Jun 9, 2009 8:21 AM PDT up reply actions 0 recs
TZ I really like the way you are experimenting with expressing stats visually
I loved the z’s you had at FD and the way you expressed an anatomy of a possession here. I agree with G, it would be interesting to see the season end game boxes so we could compare them against the King’s or a couple of player comparison boxes. I never understood the outpouring of stat hate we’ve seen here but o h well. Great work!
"We are in the business of kicking butt and business is very, very good." - Charles Barkley
by Bluejohn on Jun 9, 2009 8:51 AM PDT up reply actions 0 recs
The Kings Defensive Box...
www.mancancook.net
by vfettke on Jun 9, 2009 8:16 AM PDT reply actions 8 recs
the fundamentals of the Kings offense are strong...
They are at League average or above average in all the key criteria…
And their offense ranked 25th in the League last year.
Man, the top 24 teams must be REALLY above average.
Life is every mammal's journey from very very wet to very very dry.
by Holmdel on Jun 9, 2009 8:56 AM PDT reply actions 0 recs
My guess
Is that most teams are above average in a few key areas, and below average in others. So, a high-scoring, high paced team like the Warriors would skew the scoring average higher, but would also have higher turnover rates.
Never forget: I am a complete idiot
by Exhibit G on Jun 9, 2009 9:04 AM PDT up reply actions 0 recs
Fundamentals
Not the whole picture, obviously. Shooting is biggest part of it, and as I wrote the shooting is already above average with the opportunity to get much better.
by Ziller on Jun 9, 2009 9:12 AM PDT up reply actions 0 recs
Rec'd
I must lament, that as a certifiable layman on this subject – this is easy to understand. Can grey-ing out the difference happen? I.e. the difference from average that is clearly enough seen with hash marks, be off black or off green?
Nice thinking – have to say it – outside the box – to come up with this.
by betweentheeyes on Jun 9, 2009 9:11 AM PDT reply actions 0 recs
Coaches know their X's and O's
… There will come a time when GM’s will be expected to know their Box’s and Z’s.
by Mystic on Jun 9, 2009 9:29 AM PDT reply actions 0 recs
Great stuff
But, nothing new about that.
Evil Cowtown Inc: Screwin' Suckaz over since Nineteen Eighty-Five.....
No mistakes in the tango, darling. Not like life. Simple. That's what makes the tango so great. If you make a mistake, and get all tangled up, you just tango on.....
by pookeyguru on Jun 9, 2009 9:43 AM PDT reply actions 0 recs
Really interesting
So…to put it simply, the blue and green are positive, and the red and black are negative. If you compare other teams with varying offensive ratings, does the total area of blue + green always correspond with the better offensive team? If so, how indicative is it?
by Charlieb on Jun 9, 2009 9:57 AM PDT reply actions 0 recs
Not exactly...
As the red grows (turnovers, bad), the black shrinks, so they’re not exactly both negative. I think a more exact way to look at the black is that you want the black to be as isolated in the top right corner as possible.
by AnotherStupidSN on Jun 9, 2009 11:03 AM PDT up reply actions 0 recs
...and
The green (offensive rebounds) only happen with missed shots, so green is only “positive” in so far as it fills the gap left by missed shots. Again, the green and the blue should work together to “push” the black towards the upper right corner, which will only be allowed by also shrinking the red (turnovers). In any case, it’s a tricky and complicated thing to model, and I think TZ did a fantastic job doing it.
by AnotherStupidSN on Jun 9, 2009 11:08 AM PDT up reply actions 0 recs
There's a lot to like...
First of all, I like this. It could potentially be a way to bring non-statheads into the conversation, or at least be used as a translation of the numbers for those who aren’t number-oriented.
A couple of questions/thoughts:
- Should the actual League Averages for each category be listed? I see the line there, and can guess (based on the scale values shown), but displaying the League Average might be helpful as well.
- There seems to be two ways to set the scale for each section: (a) highest value for that season, or (b) fixed. If you set it to the highest value in each category, then the boxes become instantly easier to compare on a team-to-team basis. The downside is that it becomes harder to compare them on a year-to-year basis, as the graphics won’t be using the same scaling factors. There is a strong argument for both, in my opinion, although I would guess that in the long-run a fixed scale might have more value. That said, the immediacy of knowing where a given team ranked (visually) with the league leaders in each category would be more beneficial (again) to the laymen.
- Offensive Rating . . . what does this mean? There is no reference point to give it substance. 105.5 sounds great at first but not so much, for example, if the top rating is in the 200’s. Adding a ranking along with the rating would seem to solve this, but perhaps there’s an even better solution.
- The “black” section seems out of place. It’s a void of information that detracts from the useful parts. Going back to the “scale” idea, if the full width was set to the League Leader, then that void would carry meaning because it would represent the difference between the League Leader and the Team you are evaluating. Unless there is another value given to that void, it may be detrimental to the entire box.
This isn’t to take anything away from the basic idea/concept (or even the current implementation), because I really do think it’s forward thinking. In fact, the reason I made so many comments was that it really got me thinking, and if it could improve even more that’s just good for everyone.
I’ve got some other thoughts on this, but I’ll need to simmer on them for a while to see if there’s anything worth bringing to the table. If so, I’ll definitely post them later.
Keep up the great work Z.
by smgmatt on Jun 9, 2009 10:26 AM PDT reply actions 0 recs
One more for now
Another thought on the “black” section…
- If the scale for Offensive Rebounds is set at “total possessions” for that team, then the black section would represent Defensive Rebounds by the other team, giving a visual bar for “Offensive Rebound Percentage”. Again, this would mean that the scale used for each team varied, but the comparison would be ORP instead of Total OR, so it would still carry from team-to-team.
by smgmatt on Jun 9, 2009 10:38 AM PDT up reply actions 0 recs
So are we going to be seeing this after each game now?
That’d be cool.
Father of the "Natt this!" movement.
by Aykis16 on Jun 9, 2009 11:07 AM PDT reply actions 0 recs
Like the concept
I don’t think the geometry of the box is all the way there. Do we have any math wizards around here who can integrate the concept of the green, blue and black a little more concisely?
Also, is the defensive box an inverse of the offensive box? Could you lay one on top of the other and easily see a winning team from a losing team? Could you line up all 30 teams boxes and easily tell whose better? (This is were the green and black visually confuse me). Overall great concept, maybe one tweak you’ve made a name for yourself in basketball stat geeks hall of fame.
by bignerd on Jun 9, 2009 11:26 AM PDT reply actions 0 recs
I also like the concept...
But the asthetics need a bit of refreshing to make it more hip. Color scheme is way too early 90’s to catch on. Reminds me of an 8-ball jacket or the inside liner of a Color Me Badd cd. I say you go with current trendy league uniform colors like grayish blue, baby blue, yellow, and the ever popular flashy orange.
Damn you Robert Horry!!!
by chupacabara on Jun 9, 2009 1:15 PM PDT reply actions 0 recs
Has worked for..
The Grizz, Jazz (uni’s seperated at birth with the Grizz), Warriors, Nuggets, Thunder, and Wiz.
Damn you Robert Horry!!!
by chupacabara on Jun 9, 2009 1:18 PM PDT up reply actions 0 recs
What does the black signify?
Why a box rather than a boring bar graph? Because a box makes it seem more profound or like there is a relationship between the different metrics, when there really isn’t? (You’d think from doing it as a box that this would suggest a clear inverse relationship between turnovers and true shooting percentage, but the truth is that the relationship would not be directly inverse, correct?) The thing about visuals is that they are supposed to hit us viscerally – so we can do some Malcolm Gladwell thinking without thinking. But my visceral reaction is “huh?”
I hate to say the Emperor’s got no clothes, but unlike the anatomy of a possession, which I thought was bright and promising, this strikes me as an ill-conceived mess.
What am I missing?
Life is every mammal's journey from very very wet to very very dry.
by Holmdel on Jun 9, 2009 1:36 PM PDT reply actions 0 recs
The "Black"
Is the black hole the Enterprise morphed into the sequel that’s coming out in 2011.
Evil Cowtown Inc: Screwin' Suckaz over since Nineteen Eighty-Five.....
No mistakes in the tango, darling. Not like life. Simple. That's what makes the tango so great. If you make a mistake, and get all tangled up, you just tango on.....
by pookeyguru on Jun 9, 2009 2:03 PM PDT up reply actions 0 recs
The basis is there...
I think it just might need some tweaking.
The black area (not clearly defined in the box model) would most closely represent the “Opponent Defensive Rebound” from the Anatomy of a Possession flowchart.
Perhaps it would be better represented as RED as well to show that it was also ending a possession, but for every tweak I make on my own, the more questions I come up with.
As you stated, there is no inverse correlation between Turnovers and TS% . . . but these can be correlated visually: the further the red creeps into the box, the less chances the team had to score. So even if a team shot at a ridiculously high TS%, they still lost a lot of scoring chances through excessive Turnovers.
I think the biggest question I have right now is what defines the height & width of the box model (and if these definitions are fixed or variable), because until you know your upper & lower limits it’s hard to know where it can be improved.
That said, I think that the basis is here for something really interesting . . . something that does invoke the visceral reaction you were looking for.
by smgmatt on Jun 9, 2009 2:28 PM PDT up reply actions 0 recs
the strength of the model is that it
isnt a plain bar graph. With a bar graph you could see that all of the elements (or series, as excel would call them) would add up to 100%, and that by having fewer turnovers you would by definition gain more rebound opportunities, shots made, etc.
But the value of the box method here is by showing how a decrease in turnovers leads to an increase in offensive rating and defensive rebounds. Likewise, higher offensive rebounding numbers can only come from two areas: taking away boards from the defense, and shooting a lower percentage. As one might surmise, an offensive rebound is not always a good thing- it could mean that you blew an easy shot.
A simple bar graph wouldn’t show the inter-activity between the four elements here, but rather just their respective values and their sums.
by misterbrister on Jun 9, 2009 2:48 PM PDT up reply actions 1 recs
Brilliant Model TZ
Though…
“Half the stuff he says goes way over my head, the other half goes way, way over."
"geez its like this site is so serious i jus wanna git mai opinyons on tha Kings out there ok" - TZ
by sroufe on Jun 9, 2009 3:01 PM PDT reply actions 0 recs
Dammit
I don’t mind a little time travel, but Lost went way overboard this past season. The rumors of them simplifying the plot were INACCURATE.
by otis29 on Jun 9, 2009 4:01 PM PDT up reply actions 0 recs
Disagree
Sorry otis, IMO this season was exceptional and I never felt like the overarching plot was confusing. Overall, I’d argue that the writers did a fine job of keeping the time travel aspect of the story from becoming overly complicated for those watching each episode…
LOST is to television what LeBron is to the NBA.
"geez its like this site is so serious i jus wanna git mai opinyons on tha Kings out there ok" - TZ
by sroufe on Jun 9, 2009 6:06 PM PDT up reply actions 0 recs
They crab dribble a lot?
Evil Cowtown Inc: Screwin' Suckaz over since Nineteen Eighty-Five.....
No mistakes in the tango, darling. Not like life. Simple. That's what makes the tango so great. If you make a mistake, and get all tangled up, you just tango on.....
by pookeyguru on Jun 9, 2009 6:28 PM PDT up reply actions 0 recs
LOL
I think the polar bears could, sure.
"geez its like this site is so serious i jus wanna git mai opinyons on tha Kings out there ok" - TZ
by sroufe on Jun 9, 2009 6:51 PM PDT up reply actions 0 recs
agree sroute
they did a wonderful job of keeping the story off balance without letting it fall apart. I am intrigued to see how they will tie it all together for the final season.
LeBron – he is going to threaten a triple double average and Cleveland is going to spend the whole season falling all over themselves to get him better pieces.
by betweentheeyes on Jun 9, 2009 9:01 PM PDT up reply actions 0 recs
All I know
Is the last few episodes made me want to drink.
by otis29 on Jun 10, 2009 3:06 PM PDT up reply actions 0 recs
Great work TZ
For true shooting, your scale seems arbitrary. What if you took the mean of the TS% in the league, or if you wanted to compare over a time period, take the mean TS% of all teams for the last 10 years or 5 years. Then you could have 2 standard deviations above and below the mean. The advantage of this approach is, it would give a better feel for how out of the ordinary the TS% was.
You could take the same approach for turnovers and offensive rebounds.
Professional Hyperbole Slayer
by ForThree on Jun 9, 2009 3:18 PM PDT reply actions 0 recs
Turnovers?
We’re gonna need a bigger box.
SACTOWN ROYALTY - Try our thick creamy shakes!
by section214 on Jun 9, 2009 4:10 PM PDT up reply actions 2 recs
No -
Just thinking outside the box.
SACTOWN ROYALTY - Try our thick creamy shakes!
by section214 on Jun 9, 2009 4:33 PM PDT up reply actions 0 recs
Is this where
You start talking about your Johnson?
by otis29 on Jun 9, 2009 4:40 PM PDT up reply actions 0 recs
My Johnson is apparently less potent than Psycho-T
It might be time for Enzyte.
SACTOWN ROYALTY - Try our thick creamy shakes!
by section214 on Jun 9, 2009 4:44 PM PDT up reply actions 0 recs
Maybe I'm misunderstanding TZ
The turnover number itself should be the percentage of possessions that become turnovers, correct? It can’t be average number of turnovers per game, because that would skew against fast paced teams.
So I’m saying, put the league average in, but also put dashed white lines for one standard deviation each direction.
Professional Hyperbole Slayer
by ForThree on Jun 9, 2009 4:46 PM PDT up reply actions 0 recs
Comments For This Post Are Closed
Secondary Sidebar
User Tools
Related FanPosts on Sactown Royalty
FanPosts
Community blog posts and discussion.
Certain photos copyright © 2009 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
