Interesting note about Memphis and rim protection. Zach Randolph is a terrible rim protector and (surprisingly) Marc Gasol isn't good either. Yet, they were the 3rd best defense overall. They have good perimeter defenders and good post defenders, but this still surprised me. Conventional wisdom is that perimeter defenders rely on big men defending the rim, but, statistically, neither Gasol or Randolph are particularly good at correcting mistakes.

The inspiration for this, btw, was Hibbert. Two years ago, he was the best rim protector in the league and Indy's defense was great as a result. Last year, he was the 3rd best rim protector behind Gobert and Bogut. The big difference between this year and last, though, was the percentage of shots he contested. Is that related to his skills, or the system in Indy?

Btw, DeAndre Jordan is a massively overrated rim protector. He doesn't contest as many shots as you'd expect for his reputation and players still shoot decently when he's there. He's below Mozgov, Robin Lopez, and a number of centers who were not on the All Defensive Team. He's better than Zaza, though--sorry Dallas. Zaza's better than Marc Gasol (and we come full circle).

Well he only got there because of big baby Rivers lobbying (read whining)

you're gonna do this all season aren't you?

He's kind of right, though. Rivers' lobbying also cost Draymond Green his DPOY award. Leonard is probably the better defender, but he missed so many games. Rudy Gobert should have made one of the defensive teams. I'm not sure Jordan would have been deserving of a 3rd team.

LeBron steals rebounds, primarily from Kevin Love:

http://nyloncalculus.com/2015/10/22/201 ... cavaliers/
http://stats.nba.com/tracking/#!/player ... EFER&dir=1

I'm sure they don't care, unless it costs him a max contract.

I ended up going too deep into NBA.com's Drive stats. I took the top 52 players (plus a few Sixers and a couple others who are high in drives/game) in terms of total drives and classified play results as good, bad, or neutral. A good play was a Field Goal, a Foul Drawn, or an Assist. A bad play was a Missed Field Goal or a Turnover. A pass that did not result in an assist was neutral. I don't have hockey assist data on drives, which would probably tell me if the play was good (i.e. hockey assist) or bad (ate up a bunch of shot clock for nothing). Anyway, here are the top players sampled for ratio of good to bad on drives. This passes my LeBron test (as in LeBron is good at it, so the stat might be worth something).

Definition of Drives: "Any touch that starts at least 20 feet of the hoop and is dribbled within 10 feet of the hoop and excludes fast breaks. Measures the total number of drives as well as the points, assists and shooting percentages on drives to the basket."

Minimum 300 drives.

1. Ty Lawson 1.91/1
2. John Wall 1.85
3. LeBron James 1.85
4. James Harden 1.82
5. Dwyane Wade 1.78
6. Chris Paul 1.68
7. Jeremy Lin 1.68
8. Goran Dragic 1.67
9. Tony Parker 1.65
10. George Hill 1.65

The most notable thing about this list besides how unsurprising most of the names are is Jeremy Lin. He turns the ball over a bit too much on drives, but he draws fouls and dishes assists. His FG% is meh, but so is Ty Lawson's.

The top 10 guys who should stop driving to the basket (min = 300):
1. Trey Burke 0.90
2. Wilson Chandler 0.91
3. Josh Smith 1.03
4. Rudy Gay 1.06
5. Soloman Hill 1.06
6. Victor Oladipo 1.06
7. Brandon Knight 1.07
8. Ish Smith 1.10
9. Tyreke Evans 1.10
10. Dion Waiters 1.10

Trey Burke had a low TO%, but he shot a tick under 34% on drives. No one who played more than half the season shot that low. I like Tyreke Evans, so I'm a bit surprised to see him this low. Only 8% of his drives became assists. Considering Anthony Davis is there, that's pretty poor. The notable thing about Rudy Gay is how high his usage is. Only 5% of his drives were neutral. Only 2% became an assist. He doesn't have the lowest Assist% (Tony Wroten), but it's pretty poor.

Btw, if you just want to know the highest percentage of "good" plays, it's this:

1. LeBron 54.1%
2. Harden 53.5%
3. Derrick Rose 52.3% (Nice)
4. Steph Curry 51.8%
5. Gordon Haywood 51.1% (Nice)
6. Joe Johnson 50.9%
7. Klay Thompson 50.5%
8. Jeremy Lin 50.2%
9. Austin Rivers 50.2% (I still this this is useful)
10. Giannis Antetokounmpo 50%
10. Kyle Lowry 50%

Edit: Six fucking hours later!

http://nyloncalculus.com/2015/10/26/spo ... ng-school/

So, RAPM can be noisy. You have to use prior-informed (past seasons) RAPM to make it better. And it helps to include Box Score information. This website has both, but unfortunately, the Box Informed RAPM starts in 2007, which is nice, but not good enough. The Prior Informed goes back to 2002 (though 2004 is better, imo). The defensive numbers just seem off in the early years. Here are the Offensive RAPM

2004:
Shaquille O'Neal
Kevin Garnett
Tim Duncan
Dirk Nowitzki
Brad Miller
Rasheed Wallace
Baron Davis
Metta World Peace
Ben Wallace
Nene Hilario

2005:
Shaquille O'Neal
Tim Duncan
Manu Ginobili
Josh Howard
Jason Collins
Baron Davis
Andrei Kirilenko
Jeff Foster
Jason Kidd
Dirk Nowitzki

2006:
Shaquille O'Neal
Dwyane Wade
Manu Ginobili
Metta World Peace
Tim Duncan
Josh Howard
Andrei Kirilenko
Ben Wallace
Shane Battier
Tayshaun Prince

2007:
LeBron James
Tim Duncan
Kevin Garnett
Manu Ginobili
Baron Davis
Dwyane Wade
Metta World Peace
Gilbert Arenas
Steve Nash
Anderson Varejao

2008:
Kevin Garnett
LeBron James
Tim Duncan
Manu Ginobili
Steve Nash
Kobe Bryant
Dirk Nowitzki
Paul Pierce
Chris Bosh
Baron Davis

2009:
LeBron James
Lamar Odom
Kevin Garnett
Dwyane Wade
Chris Paul
Rashard Lewis
Kobe Bryant
Steve Nash
Ray Allen
Yao Ming


There's definitely some incorrect credit going on around here, but you get a general sense of accuracy.

Here's the Box Score Informed

2007
Tim Duncan
LeBron James
Kevin Garnett
Dirk Nowitzki
Manu Ginobili
Dwyane Wade
Gilbert Arenas
Kobe Bryant
Steve Nash
Baron Davis

2008
Kevin Garnett
LeBron James
Manu Ginobili
Dirk Nowitzki
Kobe Bryant
Chris Paul
Paul Pierce
Tim Duncan
Chris Bosh
Chauncey Billups

2009
LeBron James
Dwyane Wade
Chris Paul
Lamar Odom
Kobe Bryant
Yao Ming
Brandon Roy
Dwight Howard
Kevin Garnett
Ray Allen

2010
LeBron James
Dwyane Wade
Kevin Durant
Dwight Howard
Kobe Bryant
Tim Duncan
Deron Williams
Manu Ginobili
Anderson Varejao (still)
Dirk Nowitzki

2011
Dirk Nowitzki
LeBron James
Manu Ginobili
Dwight Howard
Chris Paul
Dwyane Wade
Kevin Garnett
Paul Pierce
Derrick Rose
Tim Duncan

2012
LeBron James
Chris Paul
Dirk Nowitzki
Marc Gasol
Blake Griffin
Tony Parker
Manu Ginobili
James Harden
Dwyane Wade
Tim Duncan

I think it's fair to say the Box Score Informed Numbers are better, no? In case you're wondering, they do better in prediction models.

Here's a comparison (2007):
Box Score None
Tim Duncan LeBron James
LeBron James Tim Duncan
Kevin Garnett Kevin Garnett
Dirk Nowitzki Manu Ginobili
Manu Ginobili Baron Davis
Dwyane Wade Dwyane Wade
Gilbert Arenas Metta World Peace
Kobe Bryant Gilbert Arenas
Steve Nash Steve Nash
Baron Davis Anderson Varejao

A lot of overlap, but Dirk does better with BI. Anderson Varejao does worse. Duncan slightly over LeBron.

2008:
Box Not
Kevin Garnett Kevin Garnett
LeBron James LeBron James
Manu Ginobili Tim Duncan
Dirk Nowitzki Manu Ginobili
Kobe Bryant Steve Nash
Chris Paul Kobe Bryant
Paul Pierce Dirk Nowitzki
Tim Duncan Paul Pierce
Chris Bosh Chris Bosh
Chauncey Billups Baron Davis

The not arguably looks better. But Chris Paul's presence makes me lean towards the Box.

Yeah looks really noisy without the box score informed

I'm not sure why they don't re-do the numbers from 2002-2007 with box score info.

Offensive Plus/Minus tends to be noisier than defensive. So, they mix that with box scores. The result is that defensive numbers look messier.

The media warps the statistics 86% of the time to glorify professional athletes.

Despite my rantings in the NBA thread, I'm a fan of raw stats over advanced metrics. They might be useful in baseball, though. I think the numbers have their limits. I don't know about this whole WAR business.

Rebounding metrics are probably the most useful metric in basketball since the slower pace of the game coupled with the much higher shooting percentages drastically lowers the raw number of rebounds available.

You prefer raw plus/minus over the advanced adjusted version? That's what it's replacing.

Hey pgm, or anyone else, can you think of any particular sets of data you'd like to see visualized in some new way? I've been dipping my toes into d3js (http://d3js.org/) for interactive visualization at work/for fun and I'm looking for interesting new things to practice on.

I'll let you know if I can think of anything. The harder part is getting the data, I think. Maybe something with the player tracking data? There's a ton of that and it's a boring mess right now.

ESPN RPM snapshot (3/8) - Everything will be in wins (so it adjusts for minutes) as assuming that coaches see something that stats don't pick up is actually a surprisingly solid adjustment.

Point Guards:

1 Stephen Curry GS
2 Russell Westbrook OKC
3 Kyle Lowry TOR
4 Chris Paul LAC
5 John Wall WSH
6 Kemba Walker CHA
7 Ricky Rubio MIN
8 Reggie Jackson DET
9 Damian Lillard POR
10 Isaiah Thomas BOS

Shooting Guards:

1 James Harden HOU
2 Jimmy Butler CHI
3 Khris Middleton MIL
4 Klay Thompson GS
5 C.J. McCollum POR
6 Victor Oladipo ORL
7 Nicolas Batum CHA
8 Kentavious Caldwell-Pope DET
9 DeMar DeRozan TOR
10 J.R. Smith CLE

Harden's defense is bad, but his offense is still good enough to keep him over Jimmy Butler. Middleton started the year slow and is now up to 3rd. Very underrated player.

Small Forwards:

1 LeBron James CLE
2 Kawhi Leonard SA
3 Paul George IND
4 Kevin Durant OKC
5 Jae Crowder BOS
6 Gordon Hayward UTAH
7 Carmelo Anthony NY
8 Danilo Gallinari DEN
9 Jared Dudley WSH
10 Trevor Ariza HOU

Durant's RPM is higher than George, but George has played more minutes. I think Crowder's numbers are a bit inflated because he's playing very well on a balanced team and the regression doesn't know how to assign the value. But he's efficient (hits 3s and gets to the line) and he plays very good defense.

Power Forwards:

1 Draymond Green GS
2 Paul Millsap ATL
3 Kevin Love CLE
4 Anthony Davis NO
5 Chris Bosh MIA
6 Marvin Williams CHA
7 Dirk Nowitzki DAL
8 Kristaps Porzingis NY
9 Marcus Morris DET
10 Amir Johnson BOS

Surprises? Anthony Davis, despite reports to the contrary, is having a good year, but not as good as people had hoped. He's a bit reliant on teammates compared to someone like Cousins and his teammates are awful. Marvin Williams's resurgence surprised me, but everyone says he's playing well. And of course, the rookie at #8 is a pretty big deal.

Centers:

1 DeAndre Jordan LAC
2 DeMarcus Cousins SAC
3 Pau Gasol CHI
4 Andre Drummond DET
5 Mason Plumlee POR
6 Brook Lopez BKN
7 Al Horford ATL
8 Zaza Pachulia DAL
9 Ian Mahinmi IND
10 Nikola Jokic DEN

I don't know if Jordan is a great center, but I do know the Jordan-Paul combination is devastating. Maybe Drummond could replace that, but Drummond's defense hasn't been as good as Jordan's this year. This is the first year where defensive hype around DAJ is probably justified.

Overall:

1 Stephen Curry, PG GS
2 Russell Westbrook, PG OKC
3 Draymond Green, PF GS
4 LeBron James, SF CLE
5 Kawhi Leonard, SF SA
6 Kyle Lowry, PG TOR
7 Chris Paul, PG LAC
8 James Harden, SG HOU
9 Paul George, SF IND
10 Kevin Durant, SF OKC

Good job skinny Lowry.

Top 10 ORPM:

1 Stephen Curry, PG GS
2 Russell Westbrook, PG OKC
3 James Harden, SG HOU
4 Chris Paul, PG LAC
5 LeBron James, SF CLE
6 Kevin Durant, SF OKC
7 Kyle Lowry, PG TOR
8 Damian Lillard, PG POR
9 Kawhi Leonard, SF SA
10 Reggie Jackson, PG DET

Top 10 DRPM:

1 Tim Duncan, C SA
2 Draymond Green, PF GS
3 Andrew Bogut, C GS
4 DeAndre Jordan, C LAC
5 Ian Mahinmi, C IND
6 Kawhi Leonard, SF SA
7 Bismack Biyombo, C TOR
8 Kevin Garnett, PF MIN
9 Andre Drummond, C DET
10 Tony Allen, SG MEM


I think RPM does a height or position adjustment where they slightly knock down centers on offense and boost them on defense, but they don't release the methodology.

So went to the Mavs-Clippers game last night. Not sure I can cosign the DeAndre defense train. His group defense looked pretty solid but he wondered at times. He also got embarrassed on post ups by David Lee one on one multiple times and multiple times Dirk made him look pretty silly.

With that said the Mavs had no legs and missed a ridiculously large number of open shots. DeAndre had several absolute wow dunks and put backs and was a really effective screener (so many guys died on his screens).

Also the issue that I see coming through on that list remains that it can't really account for role and surrounding teammates.