RealGM

Just picked up the 94-95 Sixers media guide for the dunks section, was pleasantly surprised that these were included as well (from the prior season, obviously):

http://i60.tinypic.com/15fnozr.jpg
http://i59.tinypic.com/8z21ig.jpg

Unfortunately, the 95-96 and 96-97 don't contain these sections. Evidently, after this book was released, the two guides split up: http://www.apbr.org/forum/viewtopic.php?t=2050

I'm going to try and get ahold of those two Pollack's Statistical Yearbooks (since digitalized play-by-play first exists for 96-97, these three seasons should be new data).

Anyhow though, even if raw +/- has limited utility, new data is always interesting.

EDIT - Books from Mr. Pollack arrived:

94-95:

http://i59.tinypic.com/122k8js.jpg
http://i60.tinypic.com/2e52ozo.jpg

95-96

http://i61.tinypic.com/312jwar.jpg
http://i61.tinypic.com/2vs47km.jpg

Enjoy!

I'm not sure why this interests me, but I see Terry Mills at -410 for the year. Wasn't he one of the leaders in RAPM in 1997?

His 3-point shooting got a lot better (shorter line?), and maybe he improved other facets of his game from 1994 to 1997. Plus his team got better, especially with prime Grant Hill.

I always thought Nate McMillan had some of the craziest box score stats ever. 6 points, 5 assists, 4 rebounds, 3 steals in 26 minutes per game while shooting 39 percent from 3. He led the league in steals playing nearly half the game!

I'm curious about who was tracking that stat? Was it from a play by play file or was it done by box score? Did someone watch every game after the season and recorded plus/minus?

colts18 wrote:I'm curious about who was tracking that stat? Was it from a play by play file or was it done by box score? Did someone watch every game after the season and recorded plus/minus?

According to nba.com:

During the Sixers’ offseason, Pollack and his interns analyze 1,189 play-by-play sheets from the previous season to produce Harvey Pollack’s NBA Statistical Yearbook, a 204-pagecornucopia of rare basketball information that takes statistical analysis to a new level. Want to find out the distance of every field goal taken by every player in the league from last season? Ever wonder which players get their shots blocked the most? And every year Pollack raises the bar by adding a few more nuggets of information to his book.

source: http://www.nba.com/history/pollack_hof.html

Here is the plus/minus per 100 possessions for a few stars that season. I used teams pace to estimate the # of possessions. Not perfect, but it works well.



Robinson was killing it in plus/minus that season. The Spurs completely collapsed without him. His +19.9 is really high. Only a few players in recent years have surpassed that (like 03/04 KG and 09 LeBron)

colts18 wrote:Here is the plus/minus per 100 possessions for a few stars that season. I used teams pace to estimate the # of possessions. Not perfect, but it works well.

Code: Select all

           On   Off   Net +/-
Robinson   9.4   -10.5   19.9
Malone     7.3   -10.3   17.6
Hakeem     7.0   -7.5   14.5
Shaq       5.5   -2.2   7.7
Stockton   6.3   -1.0   7.3
Pippen     5.4   -1.5   6.9
Ewing      9.1   2.7   6.3

Robinson was killing it in plus/minus that season. The Spurs completely collapsed without him. His +19.9 is really high. Only a few players in recent years have surpassed that (like 03/04 KG and 09 LeBron)

Robinson continues to look great...but the whole Malone/Stockton thing (again) doesn't jump off the page to you?

Using B-R's possessions, converted this into per 100 on/off for all players he provides data for. There is an inconstancy in the data (1 player listed twice), and I'm sure there is some human error in general, but nothing else stood out.

https://docs.google.com/spreadsheets/d/ ... sp=sharing

Thoughts:
-Robinson and Karl looking awesome
-Kevin Willis? Looks like lineup co-linearity is a possibility there, but still a crazy number.
-Dikembe looks good again
-Reggie looking good
-Rice too
-Drexler continues to bomb in this kind of +/- data... +0.3. Not a good sign.
-Richmond also with a bad number
-Payton -- woah. Anyone have an explanation for this? Pretty startling number

ElGee wrote:Using B-R's possessions, converted this into per 100 on/off for all players he provides data for. There is an inconstancy in the data (1 player listed twice), and I'm sure there is some human error in general, but nothing else stood out.

https://docs.google.com/spreadsheets/d/ ... sp=sharing

Thoughts:
-Robinson and Karl looking awesome
-Kevin Willis? Looks like lineup co-linearity is a possibility there, but still a crazy number.
-Dikembe looks good again
-Reggie looking good
-Rice too
-Drexler continues to bomb in this kind of +/- data... +0.3. Not a good sign.
-Richmond also with a bad number
-Payton -- woah. Anyone have an explanation for this? Pretty startling number

What stands out to me is Nate McMillan. His high plus/minus comes from his impressive +16.4 on court +/-. To put that into perspective, only Duncan and Ginobili in 2005 had a higher on court +/- since 2001. The +16.4 is tied with 08 KG's +/-

http://www.basketball-reference.com/pla ... y=diff_pts

Using B-R's possessions, converted this into per 100 on/off for all players he provides data for.

How is this conversion done?

Dipper 13 wrote:

Using B-R's possessions, converted this into per 100 on/off for all players he provides data for.

How is this conversion done?

This is how I did it and I assume that ElGee did it.

For example, here is David Robinson


The 4th column is the +/- per 48 minutes. The next column is Per 100 possessions where I did the conversion. The 94 Spurs played at a 90.1 pace (possessions per 48 minutes). So I multiplied the Per 48 min * 100 / (team pace ). 8.5 * 100 / 90.1 = +9.4 per 100 possessions.

This is my problem with +/- stats in general... Does anyone really think Willis and Malone were better than Hakeem in 1994?

My observations here:

-Horace Grant edges Pippen again (a trend from the late 90s +/-)

-Divac really could be underrated by a lot

ardee wrote:This is my problem with +/- stats in general... Does anyone really think Willis and Malone were better than Hakeem in 1994?

But the problem there is the conclusion, not the stat itself. PARTICULARLY raw +/- which needs to be taken at nothing beyond face value. Willis and Malone's teams outscore opponents to a greater degree with them on the floor than Hakeem's did). That's it, the stat itself tells us NOTHING more than that at all. Anything beyond that is commentary/hypothesis/conclusion based on each individual's interpretation.

This is actually the case with every stat, only it understandably is very difficult to take something like RAPM or even PER at face value, but that should really be a fundamental step in doing any kind of data analysis here.

colts18 wrote:The 4th column is the +/- per 48 minutes. The next column is Per 100 possessions where I did the conversion. The 94 Spurs played at a 90.1 pace (possessions per 48 minutes). So I multiplied the Per 48 min * 100 / (team pace ). 8.5 * 100 / 90.1 = +9.4 per 100 possessions.

Below is a screenshot provided by fpliii of the 1989-90 Sixers Plus/Minus.

http://i.imgur.com/fmqox74.jpg


Barkley looks good on offense, but the defense brings him down a bit. However it is not as bad as some might expect. He is basically neutral on defense this year, compared to some of the late 90's RAPM results.

+8.9 Net ORtg

+7.3 On Court ORtg (Compared to league average)


Net Overall: http://i.imgur.com/N3cbaAq.png

Offense: http://i.imgur.com/HBm7o4W.png

Defense: http://i.imgur.com/qfO0ijL.png

ElGee wrote:Using B-R's possessions, converted this into per 100 on/off for all players he provides data for. There is an inconstancy in the data (1 player listed twice), and I'm sure there is some human error in general, but nothing else stood out.

https://docs.google.com/spreadsheets/d/ ... sp=sharing

Thoughts:
-Robinson and Karl looking awesome
-Kevin Willis? Looks like lineup co-linearity is a possibility there, but still a crazy number.
-Dikembe looks good again
-Reggie looking good
-Rice too
-Drexler continues to bomb in this kind of +/- data... +0.3. Not a good sign.
-Richmond also with a bad number

Fantastic sir

colts18 wrote:What stands out to me is Nate McMillan. His high plus/minus comes from his impressive +16.4 on court +/-. To put that into perspective, only Duncan and Ginobili in 2005 had a higher on court +/- since 2001. The +16.4 is tied with 08 KG's +/-

http://www.basketball-reference.com/pla ... y=diff_pts

-Payton -- woah. Anyone have an explanation for this? Pretty startling number

Just some observations. I sorted purely by On Court +/-, of note: the top 4 are SEA players with >1000 MP; 6 of the top 7 are SEA players with >1000 MP. Payton, while out of this list, is the 7th Seattle player at #25 overall; doesn't look too poor in that regard. Haven't watched too much of the 94 Sonics but I suppose the floor's better spread for Kemp when Payton's off the floor. With a quick glance, honestly McMillan's insane TO levels seem to be the alarming standouts.

Here are the 8 listed:

ElGee wrote:Using B-R's possessions, converted this into per 100 on/off for all players he provides data for. There is an inconstancy in the data (1 player listed twice), and I'm sure there is some human error in general, but nothing else stood out.

https://docs.google.com/spreadsheets/d/ ... sp=sharing

Thoughts:
-Robinson and Karl looking awesome
-Kevin Willis? Looks like lineup co-linearity is a possibility there, but still a crazy number.
-Dikembe looks good again
-Reggie looking good
-Rice too
-Drexler continues to bomb in this kind of +/- data... +0.3. Not a good sign.
-Richmond also with a bad number
-Payton -- woah. Anyone have an explanation for this? Pretty startling number

It's not every day you get to look at a dataset that includes the likes of Brad Lohaus and Gary Grant

Nice work, though. Thanks!

Thanks to Elgee and Fplii, I was able to use that data to create a regressed version of RAPM.

I used 2014 Gotbuckets RAPM
Here were the inputs
2014 On per 100 possessions
2014 off per 100 possesssions
Minutes played
PER
WS/48

Using those 5 variables, I was able to come up with a formula with a R^2 value of 0.62 to 2014 Gotbuckets RAPM. Then I used those 5 variables for the 1994 players to create a regressed RAPM. The 2nd to last column is Regressed RAPM. The last column is Regressed RAPM adjusted to the variance of the 2014 Gotbuckets RAPM

https://docs.google.com/spreadsheets/d/ ... A59WS7GuA/


Top 20
1 David Robinson 5.77
2 Kevin Willis 4.30
3 Karl Malone 4.24
4 Hakeem Olajuwon 4.03
5 Nate McMillan 3.83
6 Ricky Pierce 3.27
7 Mookie Blaylock 3.25
8 Shaquille O'Neal 3.23
9 Stacey Augmon 3.12
10 Patrick Ewing 2.88
11 Dikembe Mutombo 2.87
12 Reggie Miller 2.69
13 Charles Barkley 2.60
14 John Stockton 2.59
15 Mark Price 2.49
16 Kevin Johnson 2.46
17 Shawn Kemp 2.45
18 Scottie Pippen 2.45
19 Dale Ellis 2.43
20 Horace Grant 2.30

Very cool. Initial thoughts:

-Calm down about McMillan. He played limited minutes.
-Interesting seeing more of the same from '90s Hawks players. Their rotations seem like they must have been extraordinarily simplistic under Lenny Wilkens seeing the way their players clump together.
-Malone clearly beating Stockton is really worth noting. As I've been saying, I wanted to see if Stockton's +/- trend really continued back through the S & M prime. Not looking likely.
-Ewing's raw +/- looking nothing like Hakeem or Robinson...or Mutombo.
-Miller against looking like one of the top 10 most impressive guys out there
-Payton's worth a look. Once again, a Karl star looks like someone Karl doesn't need, at least in the regular season. Don't go to crazy over Payton though. Team played great with him. All we're seeing in the net score is a relic from a limited sample size showing an OFF performance that couldn't possibly be maintained.

Doctor MJ wrote:-Malone clearly beating Stockton is really worth noting. As I've been saying, I wanted to see if Stockton's +/- trend really continued back through the S & M prime. Not looking likely.

Keep in mind it's just raw +/-. Malone also beats Stockton in raw +/- in 1997, 1998 or 1999, but RAPM tells quite different story.

Dipper 13 wrote:

colts18 wrote:The 4th column is the +/- per 48 minutes. The next column is Per 100 possessions where I did the conversion. The 94 Spurs played at a 90.1 pace (possessions per 48 minutes). So I multiplied the Per 48 min * 100 / (team pace ). 8.5 * 100 / 90.1 = +9.4 per 100 possessions.

Below is a screenshot provided by fpliii of the 1989-90 Sixers Plus/Minus.

http://i.imgur.com/fmqox74.jpg


Barkley looks good on offense, but the defense brings him down a bit. However it is not as bad as some might expect. He is basically neutral on defense this year, compared to some of the late 90's RAPM results.

+8.9 Net ORtg

+7.3 On Court ORtg (Compared to league average)


Net Overall: http://i.imgur.com/XykAtZx.png

Offense: http://i.imgur.com/HBm7o4W.png

Defense: http://i.imgur.com/qfO0ijL.png

I wonder how your net ratings of your shot charts would look if you used this method.
You used another one for them, right?