The defensive stuff is very interesting.  I noticed VAn HOrn made the
defensive stop leaders.  I assume this is an anomaly from past years
(though he masy not have been as bad a defender as he seemed when his
help D was Eschmeyer and Stephon).

On another defensive note, did you happen to chart the Mavs D in the
playoffs?  If so, was it significantly worse than in the regular
season?


--- In APBR_analysis@y..., Dean Oliver <deano@r...> wrote:
>
> Just quickly calculating some things and running them as leaders.
Here are
> leaders in terms of points produced per 100 possessions:
>
> UsedNamePossPtsProdOrtg
> Brent Barry9671193123.4
> Reggie Miller10101222121.0
> Dirk Nowitzki13921653118.7
> Chris Whitney722854118.3
> John Stockton10631253117.9
> Steve Nash13631603117.6
> Elton Brand12531470117.3
> Wesley Person8841036117.2
> Ray Allen12331422115.3
> Predrag Stojakovic11781357115.2
> Chauncey Billups9631102114.4
> Derek Fisher656751114.4
> Shaquille O'Neal14521656114.0
> Brad Miller9041030114.0
> Rashard Lewis9661093113.1
> Steve Smith749847113.1
> Tim Duncan18252053112.5
> Andre Miller14521632112.4
> David Robinson833932111.9
> Antonio Daniels681762111.9
>
> Leaders in Scoring Possessions per possession:
>
> UsedNameScPossPossFloor%
> Shaquille O'Neal85014520.585
> Dale Davis3776460.584
> Elton Brand72812530.581
> David Robinson4698330.563
> John Stockton59810630.562
> Brad Miller5069040.560
> Tim Duncan100618250.551
> Ruben Patterson3937280.540
> Andre Miller78214520.539
> Anthony Mason4378110.538
> Dikembe Mutombo4528400.538
> Dirk Nowitzki74713920.536
> Brent Barry5169670.534
> Andrei Kirilenko4057580.533
> Kobe Bryant95317960.531
> Chris Webber63511980.531
> Pau Gasol69313090.529
> Gary Payton92817540.529
> Elden Campbell4889240.528
> Donyell Marshall3877360.526
>
>
> Points allowed per 100 possessions:
>
> UsedNameDefStopsStops/PossDRtg
> Ben Wallace8610.77389.7
> David Robinson5800.65992.0
> Tim Duncan8140.64192.7
> Alonzo Mourning5600.61694.3
> Dikembe Mutombo6680.60994.4
> Malik Rose3910.59494.6
> Jason Kidd6760.56795.4
> Scot Pollard4560.59995.7
> Kenyon Martin5320.54596.3
> Elden Campbell5060.61696.3
> Paul Pierce7440.57696.5
> Vlade Divac5660.57996.5
> Keon Clark4990.60596.5
> Keith Van Horn5170.53896.6
> Tony Battie4080.57396.6
> Aaron Williams3210.53196.8
> Robert Horry4830.57896.8
> Andrei Kirilenko5180.62997.1
> Shawn Marion7380.61097.1
> Jamaal Magloire3470.58797.5
>
> Points produced per game:
>
> UsedNameGPtsProdPtsProd/G
> Allen Iverson60178129.7
> Tracy McGrady76193025.4
> Tim Duncan82205325.0
> Kobe Bryant80198424.8
> Shaquille O'Neal67165624.7
> Paul Pierce82198124.2
> Gary Payton82195823.9
> Antoine Walker81180122.2
> Jerry Stackhouse76166621.9
> Stephon Marbury82178621.8
> Dirk Nowitzki76165321.7
> Kevin Garnett81174921.6
> Karl Malone80170921.4
> Shareef Abdur-Rahim77159020.7
> Ray Allen69142220.6
> Sam Cassell74152420.6
> Andre Miller81163220.2
> Baron Davis82163519.9
> Cuttino Mobley74147319.9
> Steve Nash82160319.5
>
>
> Dean Oliver
> Journal of Basketball Studies
> http://www.rawbw.com/~deano/index.html
> deano@r...

--- alleyoop2 <alleyoop2@...> wrote:
> Anyone know where I can find first-half and
> second-half Win-Loss
> records for every team for, say, the last ten years
> or so?
>
>
STATS Inc. should have that info. It's in their yearly
handbooks.

Ed

__________________________________________________
Do You Yahoo!?
LAUNCH - Your Yahoo! Music Experience
http://launch.yahoo.com

On Fri, 17 May 2002, Ed Weiland wrote:

>
> --- alleyoop2 <alleyoop2@...> wrote:
> > Anyone know where I can find first-half and
> > second-half Win-Loss
> > records for every team for, say, the last ten years
> > or so?
> >
> >
> STATS Inc. should have that info. It's in their yearly
> handbooks.

I don't have any answers, but I'm nonetheless wondering about the
question:  do you mean first-half and second-half of a game (who scored
the most points in a half), or a season (won-loss over first 41 games, and
last 41 games)?


--MKT

For a season. The Stats handbooks list it for the previous season,
but I would need a decade's worth of handbooks to do what I want to
do, and unfortunately I don't have that.



--- In APBR_analysis@y..., "Michael K. Tamada" <tamada@o...> wrote:
>
>
> On Fri, 17 May 2002, Ed Weiland wrote:
>
> >
> > --- alleyoop2 <alleyoop2@y...> wrote:
> > > Anyone know where I can find first-half and
> > > second-half Win-Loss
> > > records for every team for, say, the last ten years
> > > or so?
> > >
> > >
> > STATS Inc. should have that info. It's in their yearly
> > handbooks.
>
> I don't have any answers, but I'm nonetheless wondering about the
> question:  do you mean first-half and second-half of a game (who
scored
> the most points in a half), or a season (won-loss over first 41
games, and
> last 41 games)?
>
>
> --MKT

I'm surprised that rebounds didn't correlate more strongly with
winning, simply because those are the games when their opponent
shoots a lower percentage and thus there are more defensive rebounds
to be grabbed.




--- In APBR_analysis@y..., "sonicdk2" <sonicdk2@y...> wrote:
> A while ago we discussed the usefulness of identifying the
predictive
> value of a statistic.  When it comes to the median, rather than the
> mean, perhaps the median would be better when a player's scores are
> highly skewed.  I have analyzed four players whose distributions
are
> NOT highly skewed--Brent Barry, Gary Payton, Vin Baker, Rashard
> Lewis, and in the future will move to the more skewed player's
> stats.
>
> While looking at these 4 players (so I could contrast them with the
> more skewed players), I noticed a few interesting things, which I
> report below.  So the skewed players will just have to wait!  I'll
> get to them eventually.
>
> Instead of comparing to see if the mean or median provides better
> predictive value, I looked to see if individual stats had any
> predictive value in the first place.  In regard to predictive
value,
> I decided to try a tough test--do they predict whether a team will
> win or lose?  When a player scores more than average, has he
improved
> his team's chance of winning?  When he scores less than average,
does
> this lower his team's chance of winning?  Here are my findings for
> points, rebounds, and assists, from the 2001-2002 regular season.
>
> I did not use linear regression because I didn't want to assume
there
> would be a linear relation between variables.  I did not use
> nonlinear regression either because I wanted a simpler method.
>
> If you want the excruciating details of how I arrived at my
> conclusions, please let me know.  I will gladly send you a much
much
> more detailed description of my method. :)
>
> 1.)  When Brent Barry scored more than his average, the Sonics'
> chance of winning went up by more than 26%.  When he had more
assists
> than average, his team's chance of winning went up by more than
30%.
> An application of this trend is the recent playoff series--when the
> Spurs stopped Brent Barry from scoring and passing, it really hurt
> the Sonics.  His rebounding had no major impact on winning or
losing
> during the regular season, even though his average is good for a
> guard.
>
> 2.) Gary Payton, not surprisingly, improved his team's chance of
> winning by about 18% when he got more than his average number of
> assists.  His scoring and rebounding may be important, but
variations
> in them did not make a difference in the Sonics' winning or
losing.
> This might be due to GP's ability to do other things to make up for
> when he's not scoring or rebounding.  Of course, this data is all
> nonexperimental so there might be other factors responsible for the
> presence or absence of a relationship of one stat with winning %.
>
> 3.) Variations in Vin Baker's stats (points, rebounds, assists) had
> no appreciable relation to the Sonics' winning or losing.
>
> 4.) Although there were no significant results for Rashard Lewis,
the
> Sonics seemed to win a little bit more when  Rashard got fewer
> rebounds or fewer assists than average.  Maybe a fluke.
>
> My main conclusion is that it looks useful to look at which
> individual player stat might show predictive value for a team's
> winning or losing--it can help an opponent figure out how to defend
a
> player, for one thing.  Of course, combinations of various player
> stats would presumably have better predictive validity, such as Pts
+
> Rebs + Asts - TOs etc.
>
> Player     Stat     Mean  Change in Winning % when Stat is Above
Mean
> Barry     Points   14.37    +26.7%*
>           Rebs      5.42     +5.4%
>           Assists   5.26    +30.4%*
> Payton    Points    22.13   +10.7%
>           Rebs      4.82     +0%
>           Assists   8.99    +18.9%*
> VBaker    Points    14.07   +0.9%
>           Rebs       6.36   +3.5%
>           Assists    1.31   -5.8%
> RLewis    Points    16.83   +7.2%
>           Rebs       7.01   -8.3%
>           Assists    1.73   -9.0%
> (Note:  * means statistically significant at alpha of .05 per test.)
>
> Cutoff values for significance:
> Barry W-L change:  +/-15.34%
> Gary Payton W-L change: +/-15.23%
> Vin Baker W-L change:  +/-18.66%
> R Lewis W-L change:  +/-11.54%
>
> Comments?
>
> Dennis Keefe

Prof. Berri's work indicated that offensive rebounds had the highest correlation
to wins, followed by three-pointers made. A defensive rebound just wasn't that
big a deal.

One interesting thing from the recently completed Laker-Spurs series is that the
Lakers KILLED them with second chance points. While San Antonio did okay with
offensive rebounding they simply couldn't convert those second chances into
points. I'll have to add my stuff up, but the Spurs were well under 1 point per
second chance opportunity while the Lakers were near 1.5 points per second
chance opportunity.

-----Original Message-----
From: alleyoop2@...
Sent: Friday, May 17, 2002 11:11 AM
To: alleyoop2@...; APBR_analysis@yahoogroups.com
Subject: [APBR_analysis] Re: Predicting Wins and Losses From a Single
Player Stat


I'm surprised that rebounds didn't correlate more strongly with
winning, simply because those are the games when their opponent
shoots a lower percentage and thus there are more defensive rebounds
to be grabbed.




--- In APBR_analysis@y..., "sonicdk2" <sonicdk2@y...> wrote:
> A while ago we discussed the usefulness of identifying the
predictive
> value of a statistic.  When it comes to the median, rather than the
> mean, perhaps the median would be better when a player's scores are
> highly skewed.  I have analyzed four players whose distributions
are
> NOT highly skewed--Brent Barry, Gary Payton, Vin Baker, Rashard
> Lewis, and in the future will move to the more skewed player's
> stats.
>
> While looking at these 4 players (so I could contrast them with the
> more skewed players), I noticed a few interesting things, which I
> report below.  So the skewed players will just have to wait!  I'll
> get to them eventually.
>
> Instead of comparing to see if the mean or median provides better
> predictive value, I looked to see if individual stats had any
> predictive value in the first place.  In regard to predictive
value,
> I decided to try a tough test--do they predict whether a team will
> win or lose?  When a player scores more than average, has he
improved
> his team's chance of winning?  When he scores less than average,
does
> this lower his team's chance of winning?  Here are my findings for
> points, rebounds, and assists, from the 2001-2002 regular season.
>
> I did not use linear regression because I didn't want to assume
there
> would be a linear relation between variables.  I did not use
> nonlinear regression either because I wanted a simpler method.
>
> If you want the excruciating details of how I arrived at my
> conclusions, please let me know.  I will gladly send you a much
much
> more detailed description of my method. :)
>
> 1.)  When Brent Barry scored more than his average, the Sonics'
> chance of winning went up by more than 26%.  When he had more
assists
> than average, his team's chance of winning went up by more than
30%.
> An application of this trend is the recent playoff series--when the
> Spurs stopped Brent Barry from scoring and passing, it really hurt
> the Sonics.  His rebounding had no major impact on winning or
losing
> during the regular season, even though his average is good for a
> guard.
>
> 2.) Gary Payton, not surprisingly, improved his team's chance of
> winning by about 18% when he got more than his average number of
> assists.  His scoring and rebounding may be important, but
variations
> in them did not make a difference in the Sonics' winning or
losing.
> This might be due to GP's ability to do other things to make up for
> when he's not scoring or rebounding.  Of course, this data is all
> nonexperimental so there might be other factors responsible for the
> presence or absence of a relationship of one stat with winning %.
>
> 3.) Variations in Vin Baker's stats (points, rebounds, assists) had
> no appreciable relation to the Sonics' winning or losing.
>
> 4.) Although there were no significant results for Rashard Lewis,
the
> Sonics seemed to win a little bit more when  Rashard got fewer
> rebounds or fewer assists than average.  Maybe a fluke.
>
> My main conclusion is that it looks useful to look at which
> individual player stat might show predictive value for a team's
> winning or losing--it can help an opponent figure out how to defend
a
> player, for one thing.  Of course, combinations of various player
> stats would presumably have better predictive validity, such as Pts
+
> Rebs + Asts - TOs etc.
>
> Player     Stat     Mean  Change in Winning % when Stat is Above
Mean
> Barry     Points   14.37    +26.7%*
>           Rebs      5.42     +5.4%
>           Assists   5.26    +30.4%*
> Payton    Points    22.13   +10.7%
>           Rebs      4.82     +0%
>           Assists   8.99    +18.9%*
> VBaker    Points    14.07   +0.9%
>           Rebs       6.36   +3.5%
>           Assists    1.31   -5.8%
> RLewis    Points    16.83   +7.2%
>           Rebs       7.01   -8.3%
>           Assists    1.73   -9.0%
> (Note:  * means statistically significant at alpha of .05 per test.)
>
> Cutoff values for significance:
> Barry W-L change:  +/-15.34%
> Gary Payton W-L change: +/-15.23%
> Vin Baker W-L change:  +/-18.66%
> R Lewis W-L change:  +/-11.54%
>
> Comments?
>
> Dennis Keefe



To unsubscribe from this group, send an email to:
APBR_analysis-unsubscribe@yahoogroups.com



Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

--- Ed Weiland <weiland1029@...> wrote:
>
> --- alleyoop2 <alleyoop2@...> wrote:
> > Anyone know where I can find first-half and
> > second-half Win-Loss
> > records for every team for, say, the last ten
> years
> > or so?
> >
> >
> STATS Inc. should have that info. It's in their
> yearly
> handbooks.
>

If you can't find it quickly I can do it this weekend.
  Let me know if you want me to generate this.

DeanL

> __________________________________________________
> Do You Yahoo!?
> LAUNCH - Your Yahoo! Music Experience
> http://launch.yahoo.com
>


__________________________________________________
Do You Yahoo!?
LAUNCH - Your Yahoo! Music Experience
http://launch.yahoo.com

--- In APBR_analysis@y..., "harlanzo" <harlanzo@y...> wrote:
> The defensive stuff is very interesting.  I noticed VAn HOrn made
the
> defensive stop leaders.  I assume this is an anomaly from past
years
> (though he masy not have been as bad a defender as he seemed when
his
> help D was Eschmeyer and Stephon).
>

Van Horn's number is better than in year's past.  Everyone on the
Nets improved substantially this year.  Defense is such a team thing,
guys working together.  Individual defense is hard to measure.  I'm
thinking that there really are only a few players in the league who
truly can dictate a team defense.  Most go along with the scheme and
are improved/hurt by the level of communication/coordination.

As I've mentioned before, we're undertaking an effort in the WNBA to
take some additional stats to help characterize defense.  Things like
field goals allowed, field goals stopped (not necessarily blocked),
forced turnovers (not necessarily steals).  This won't get
everything.  Here are 2 examples.  1 guy shows 10 stops and 11 scores
allowed.  Another guy shows 5 stops and 5 scores allowed.  Who is the
better defender?  What if the first guy was covering someone who
normally scores on 15 of 25 possessions and the second guy was
covering someone who usually scores on 4 of 9 possession.

Because defense can match up against the offense, who guys match up
against can be an important factor.  We're not trying for that here.
Doug Steele tries with mixed results.


> On another defensive note, did you happen to chart the Mavs D in
the
> playoffs?  If so, was it significantly worse than in the regular
> season?

I have the numbers.  Give me a day or two to look at them.  Keep in
mind that they did clean Minnesota's clock.  Or are you just
interested in the series against Sacramento?  I'm guessing you only
remember the last series.

I'll look.

DeanO

--- In APBR_analysis@y..., "sonicdk2" <sonicdk2@y...> wrote:
> Instead of comparing to see if the mean or median provides better
> predictive value, I looked to see if individual stats had any
> predictive value in the first place.  In regard to predictive
value,
> I decided to try a tough test--do they predict whether a team will
> win or lose?  When a player scores more than average, has he
improved
> his team's chance of winning?  When he scores less than average,
does
> this lower his team's chance of winning?  Here are my findings for
> points, rebounds, and assists, from the 2001-2002 regular season.
>

This is a good approach.  What is interesting, though, is that a
measure of a team's quality is when such trends CANNOT be found.
That means that they are robust to strategy.  San Antonio could put
Bowen on Barry because the Sonics aren't robust to that strategy.
Hack-a-Shaq doesn't particularly help or hurt a team significantly --
the Lakers are robust to that strategy.

Teams look for these little trends themselves and try to take
advantage of them.  I have a huge set of programs designed to look
for strategies to beat teams.  A lot of times, the trends just don't
exist.

One of my favorite from when I scouted regularly -- does this team
like a tightly called game or a loosely called game?  I absolutely
hated answering this question!  But there are a few teams every so
often where it matters.  For most, it doesn't.

>
> Player     Stat     Mean  Change in Winning % when Stat is Above
Mean
> Barry     Points   14.37    +26.7%*
>           Rebs      5.42     +5.4%
>           Assists   5.26    +30.4%*
> Payton    Points    22.13   +10.7%
>           Rebs      4.82     +0%
>           Assists   8.99    +18.9%*
> VBaker    Points    14.07   +0.9%
>           Rebs       6.36   +3.5%
>           Assists    1.31   -5.8%
> RLewis    Points    16.83   +7.2%
>           Rebs       7.01   -8.3%
>           Assists    1.73   -9.0%
> (Note:  * means statistically significant at alpha of .05 per test.)

Glad you looked at statistical significance.  It makes a big
difference to do so.  I see a lot of little trends that disappear if
you don't look at significance.

As to why rebounds are not significant -- I'm still looking for a way
to explain this.  I know Berri's results and I just don't quite trust
the way he applied them to individuals.  Sure, a regression on teams
shows offensive rebounds are important.  I don't doubt it.  But there
are a lot of guys who get tons of OR's who just aren't valuable.
Somehow Berri's method overvalued OR's on an individual scale.  Berri
said in the paper that different positions on the court are
irreplaceable (or something like that), which is why he scaled
against position.  That's how the average wins per position should
end up the same.  First of all, that was speculation.  But second, it
seems like the roles players have is not defined by their position as
much as their shooting vs. rebounding vs. passing numbers.

I really dunno.

The Seattle perspective:

--- In APBR_analysis@y..., "sonicdk2" <sonicdk2@y...> wrote:
> I have analyzed four players whose distributions are NOT highly
> skewed--Brent Barry, Gary Payton, Vin Baker, Rashard Lewis, and in
> the future will move to the more skewed player's stats.

Interesting. One of the myriad projects that lies on my mental
checklist is going through the Sonics roster's scoring distribution
to try to see what level of inconsistency each player has.

> 1.)  When Brent Barry scored more than his average, the Sonics'
> chance of winning went up by more than 26%.  When he had more
> assists than average, his team's chance of winning went up by more
> than 30%.

Yeah, this doesn't come as much surprise. I'm sure I've remarked on
at least one occasion, "As Brent Barry goes, so go the Sonics." The
first-round series neatly illustrated it. Deciding if his problems
were because of Bruce Bowen or his own dropoff is something I wish
we could do with stats.

> 2.) Gary Payton . . .His scoring and rebounding may be important,
> but variations in them did not make a difference in the Sonics'
> winning or losing.

I'd say the main reason for this, at least the scoring, might be
that Payton is a rather consistent scorer. I don't believe he was in
single-digits more than once or twice this season, but his peak
scoring is not as high as many players. He was only over 30 a
handful of times, with a real outlier in his 43-point effort against
the Clippers. I wish I had the file on this handy, but it's at
school.

> 3.) Variations in Vin Baker's stats (points, rebounds, assists)
> had no appreciable relation to the Sonics' winning or losing.

I *think* that this was what I found a few months ago when I was
looking at something similar. I'm still trying to figure out why
giving him more possessions helped the team. . . .

> 4.) Although there were no significant results for Rashard Lewis,
> the Sonics seemed to win a little bit more when  Rashard got fewer
> rebounds or fewer assists than average.  Maybe a fluke.

I think a possibility here is that Lewis' rebounding was much better
at the beginning of the season, when the team was struggling a bit.
His play declined, but this was more than made up for by
improvements in several other players' play.

> My main conclusion is that it looks useful to look at which
> individual player stat might show predictive value for a team's
> winning or losing--it can help an opponent figure out how to
> defend a player, for one thing.

An excellent application, one I had never considered. But logically,
it made sense in the first-round series with San Antonio. Payton,
defended by a much weaker defender in Parker, still never really
went off. Barry, defended by Bowen, did absolutely nothing until the
fourth quarter of game five. That could have been the difference for
the Spurs.

One question that this has to raise, however, is to what extent
variations in a players' numbers are of their own making. We are
looking at a number of extraneous variables, such as the combination
of teammates and the team's success. For example, the Sonics were
rarely blown out in the season's second half while putting a few
whoopins on opponents. A blowout lowers minutes, which should lower
the accumulation of points, etc. So the Sonics' starters should have
lower production in blowouts, which are disproportionately
victories. This could dilute the true effect of these starters'
effectiveness on the Sonics performance.

And the question raised by Lewis' rebounding performance in
November, as I see it, is whether this is the product of the fact
that the Sonics were giving heavy minutes to Lewis at the power
forward and Baker (a mediocre rebounder) at the center. In that
lineup, Lewis is bound to accumulate more rebounds despite not
actually having 'improved' his own play.

All in all, an excellent effort. Any chance of turning it into a
guest column at SonicsCentral?

--- In APBR_analysis@y..., Dean Oliver <deano@r...> wrote:
>
> Who does Byron Scott have over there coaching defense?
>
It is pretty much the same staff as last year.  Eddie Jordan is
the "defensive specialist."

--- In APBR_analysis@y..., "thedawgsareout" <kpelton08@h...> wrote:
> The Seattle perspective:
> that Payton is a rather consistent scorer. I don't believe he was
in
> single-digits more than once or twice this season, but his peak
> scoring is not as high as many players. He was only over 30 a
> handful of times, with a real outlier in his 43-point effort
against
> the Clippers.
Yup.  Lowest was 11 points (twice), highest was 32 (3 times) rexcept
for that outlier.  GP's standard deviation in scoring was 5.427
points, which is pretty low.


looking at a number of extraneous variables, such as the combination
> of teammates and the team's success.
multiple regression on win/loss might help--put minutes played or
margin of victory/loss as first predictor and then the stat of
interest such as points scored as second predictor and see what you
get.  But I'm not sure how to carry it out properly since the
predictor variables are likely to be correlated, which tends to
invalidate the result.  Anybody have any ideas?
>


  Any chance of turning it into a
> guest column at SonicsCentral?
I dunno.  If you get a draft of a guest column within the next week,
then I guess you'll know....

Here's an application of splitting the win/loss record into games
above and below a player's mean.  I did it for only the stars of the
Nets-Celtics game.

Paul Pierce:
points scored +13.6%*
(in other words, Celtics are 13.6% more likely to win when Paul
scores above his average.)
offensive rebounds -5.5%
defensive rebounds -10.5% (?)
Assists: +2.0%

* indicates statistical significance at .05 level per test.

The result for drebs was near significance (10.7% was cutoff for
significance), and splitting by the median rather than the mean
yielded significance (-13.3%).  So Kenyon, with his subpar
rebounding, should let Paul get those defensive boards, then give the
celtics a taste of their own medicine:  set a screen on him so he has
trouble getting down the court (should sound familiar to those
reading the APBR site posts).  Thus Paul won't get downcourt very
fast and the offense will miss their #1 option, and maybe he'll score
less.


ANTOINE WALKER
points:  -2.7%
off rebounds:  -4.2%
def rebounds: -3.5%
assists:  +38.8%*

cutoff for significance again is 10.7%

So obviously van horn should key on getting walker to shoot rather
than pass.

JASON KIDD
points:  +12.8%*
off rebs:  +5.8%
def rebs: +6.0%
assists:  +16.7%*

cutoff for significance:  10.5%

Keep Jason from scoring.  Good luck trying to stop him from passing
to the open man!


dennis keefe

--- In APBR_analysis@y..., "sonicdk2" <sonicdk2@y...> wrote:
> Here's an application of splitting the win/loss record into games
> above and below a player's mean.  I did it for only the stars of
the
> Nets-Celtics game.
>
> Paul Pierce:
> points scored +13.6%*
> (in other words, Celtics are 13.6% more likely to win when Paul
> scores above his average.)
> offensive rebounds -5.5%
> defensive rebounds -10.5% (?)
> Assists: +2.0%
>
> * indicates statistical significance at .05 level per test.
>
> The result for drebs was near significance (10.7% was cutoff for
> significance), and splitting by the median rather than the mean
> yielded significance (-13.3%).  So Kenyon, with his subpar
> rebounding, should let Paul get those defensive boards, then give
the
> celtics a taste of their own medicine:  set a screen on him so he
has
> trouble getting down the court (should sound familiar to those
> reading the APBR site posts).  Thus Paul won't get downcourt very
> fast and the offense will miss their #1 option, and maybe he'll
score
> less.
>

Actually, an interesting thing about both conference finals is that
they both feature teams that rely on the refs.  Boston needs fouls to
be called tightly to win games.  The Lakers rely on refs not calling
much against Shaq down low.  Given that refs usually swallow their
whistles in the playoffs, this gives a rather substantial advantage
to both Jersey and the Lakers.  Generally, I don't like series where
you have teams of such great contrast in that type of style.  It
means that there will be a lot of complaining about the refs, which
there was in yesterday's Sac-LA game.

If Pierce can draw fouls, the Celtics have a very good chance.

>
> ANTOINE WALKER
> points:  -2.7%
> off rebounds:  -4.2%
> def rebounds: -3.5%
> assists:  +38.8%*
>
> cutoff for significance again is 10.7%
>
> So obviously van horn should key on getting walker to shoot rather
> than pass.

I think I found the same thing a few years ago.  Walker gets a little
excited when he makes 2 in a row, then thinks he can make anything
from anywhere with anyone on him.

>
> JASON KIDD
> points:  +12.8%*
> off rebs:  +5.8%
> def rebs: +6.0%
> assists:  +16.7%*
>
> cutoff for significance:  10.5%
>
> Keep Jason from scoring.  Good luck trying to stop him from passing
> to the open man!

I've generally found that the Nets offense is pretty robust.  It's
not a very efficient offense, but it's pretty robust to strategy
variations.  I really found just one reasonable statistical indicator
for them -- keep them from getting high percentage 3 pt shots.  They
don't shoot a lot of them, but try to keep them from being good
ones.  In attacking them, you need to get to the basket and draw
fouls.  Boston standing at the 3pt line without driving probably
won't win.  Though not nearly as strong an indicator as it is for
Boston and the Lakers, the Nets don't like tightly called games.

I noticed most of the guys in this study have a high correlation
between assists and winning, and I think cause and effect may be
backwards here. Assists only result from made field goals; when you
win you normally make more field goals than the other team, and
normally that means you'll end up with more assists.


--- In APBR_analysis@y..., "HoopStudies" <deano@r...> wrote:
> --- In APBR_analysis@y..., "sonicdk2" <sonicdk2@y...> wrote:
> > Here's an application of splitting the win/loss record into games
> > above and below a player's mean.  I did it for only the stars of
> the
> > Nets-Celtics game.
> >
> > Paul Pierce:
> > points scored +13.6%*
> > (in other words, Celtics are 13.6% more likely to win when Paul
> > scores above his average.)
> > offensive rebounds -5.5%
> > defensive rebounds -10.5% (?)
> > Assists: +2.0%
> >
> > * indicates statistical significance at .05 level per test.
> >
> > The result for drebs was near significance (10.7% was cutoff for
> > significance), and splitting by the median rather than the mean
> > yielded significance (-13.3%).  So Kenyon, with his subpar
> > rebounding, should let Paul get those defensive boards, then give
> the
> > celtics a taste of their own medicine:  set a screen on him so he
> has
> > trouble getting down the court (should sound familiar to those
> > reading the APBR site posts).  Thus Paul won't get downcourt very
> > fast and the offense will miss their #1 option, and maybe he'll
> score
> > less.
> >
>
> Actually, an interesting thing about both conference finals is that
> they both feature teams that rely on the refs.  Boston needs fouls
to
> be called tightly to win games.  The Lakers rely on refs not
calling
> much against Shaq down low.  Given that refs usually swallow their
> whistles in the playoffs, this gives a rather substantial advantage
> to both Jersey and the Lakers.  Generally, I don't like series
where
> you have teams of such great contrast in that type of style.  It
> means that there will be a lot of complaining about the refs, which
> there was in yesterday's Sac-LA game.
>
> If Pierce can draw fouls, the Celtics have a very good chance.
>
> >
> > ANTOINE WALKER
> > points:  -2.7%
> > off rebounds:  -4.2%
> > def rebounds: -3.5%
> > assists:  +38.8%*
> >
> > cutoff for significance again is 10.7%
> >
> > So obviously van horn should key on getting walker to shoot
rather
> > than pass.
>
> I think I found the same thing a few years ago.  Walker gets a
little
> excited when he makes 2 in a row, then thinks he can make anything
> from anywhere with anyone on him.
>
> >
> > JASON KIDD
> > points:  +12.8%*
> > off rebs:  +5.8%
> > def rebs: +6.0%
> > assists:  +16.7%*
> >
> > cutoff for significance:  10.5%
> >
> > Keep Jason from scoring.  Good luck trying to stop him from
passing
> > to the open man!
>
> I've generally found that the Nets offense is pretty robust.  It's
> not a very efficient offense, but it's pretty robust to strategy
> variations.  I really found just one reasonable statistical
indicator
> for them -- keep them from getting high percentage 3 pt shots.
They
> don't shoot a lot of them, but try to keep them from being good
> ones.  In attacking them, you need to get to the basket and draw
> fouls.  Boston standing at the 3pt line without driving probably
> won't win.  Though not nearly as strong an indicator as it is for
> Boston and the Lakers, the Nets don't like tightly called games.

--- In APBR_analysis@y..., "alleyoop2" <alleyoop2@y...> wrote:
> I noticed most of the guys in this study have a high correlation
> between assists and winning, and I think cause and effect may be
> backwards here. Assists only result from made field goals; when you
> win you normally make more field goals than the other team, and
> normally that means you'll end up with more assists.

True, in a sense I agree with your assessment of cause an effect, but
assists are also an indication of the manner by which team points are
scored (rather than creating one's own shot, or by free throws.) My
main focus with assists is to see who an opponent should focus on
defensively if they want to make sure the opponent doesn't make too
many timely passes to the open man.

Other players show weaker relations of assists with winning, so it's
not as important to shut down that aspect of their game--such as guys
who like to keep the ball, looking for a shot, and then pass at the
last few seconds--they might get an assist but it's not a great
winning strategy and so the passing lanes don't need to be covered so
tightly with such a player.  Am I making sense?

Change the way the opponent plays and that may put them at a
disadvantage.

--- HoopStudies <deano@...> wrote:
> --- In APBR_analysis@y..., "Michael K. Tamada"
> <tamada@o...> wrote:
> >
> We really need a Project Play-By-Play, where we
> develop a db of play-
> by-plays.  I am willing to volunteer a bunch of
> play-by-plays.  I can
> probably get most of them into electronic format,
> though some will be
> in different formats than others.  Who wants to
> develop the db
> structure and import things?
>

I'm willing to look at it and give an estimate of the
effort involved.

DeanL



__________________________________________________
Do You Yahoo!?
LAUNCH - Your Yahoo! Music Experience
http://launch.yahoo.com

Back in the early '90s, when Dennis Rodman decided to be a rebound
specialist, I thought I noticed a subsequent decline in his
teammates' rebounding numbers.  Laimbeer was actually in the twilight
of his career, so his rebounding rate might be expected to fade.

Upon the Worm's move to San Antonio, David Robinson's career peaked,
in every measurable way except rebounding.  Same effect?  Not that
it's surprising, but I wonder if overall team rebounds are helped by
a Rodman type; or if other categories are boosted at the same time;
etc.

I found 66 players who played significant minutes both with and
without Rodman, in Detroit, San Antone, and Chicago.  I took a simple
ratio of their rebounding rate (with/without the worm), as well as
other production rates.  Standardized, of course.

Broken down by city:

Tem Min. Pct. Sco. Reb. Ast. P.F. Stl. (TO) Blk. - Total

Det 0.85 1.00 0.99 0.85 0.86 1.01 0.86 0.87 0.88 - 0.96

SAn 0.93 1.01 1.05 0.88 0.99 0.97 1.06 0.95 1.01 - 1.00

Chi 0.78 1.01 1.04 0.83 1.09 1.03 0.77 0.91 1.19 - 0.99


Now by position:

Pos Min. Pct. Sco. Reb. Ast. P.F. Stl. (TO) Blk. - Total

C - 0.79 0.97 1.00 0.85 1.16 1.01 0.72 1.01 1.01 - 0.96

F - 0.88 1.04 1.05 0.80 0.90 1.05 0.89 0.89 0.89 - 0.97

G - 0.95 1.03 1.05 0.91 1.00 1.01 1.01 0.82 1.10 - 1.04


Curiously, Rodman tended to cut into centers' minutes more than into
forwards'.  Then again, the Spurs' backup centers were J.R. Reid and
Antoine Carr, both really forwards.  Terry Cummings, too.  And the
Bulls ran without centers a lot, when Rodman was there.

Rodman's rising assist-enhancement rate seems to suggest he learned
more nuances of offense, without being the actual ball-distributor?

The data used in the ratios above were for whole player-seasons with
and without Rodman as a teammate.


Mike G

Funny, I woke up thinking about Rodman this morning.  That's actually
not funny.  Trust me.  But it's funny that MikeG did something on
Rodman, too.

Dave Berri, as some of you remember, had his evaluation method that
ranked Rodman the best player in the league, creating more net wins
than any other player in that year in Chicago.  Berri also has said
that good players have a negative effect on others by doing an
analysis somewhat like MikeG did here.  MikeG doesn't say Rodman has
a negative effect, which is good.  I think what we are looking at
replacement of finite number of rebounds.  Robinson didn't get less
valuable when Rodman came though his rebounds declined.  Not sure
what Berri's numbers show.  The debate on Rodman's value will last
for a long time.  It will keep him out of the Hall of Fame a while, I
think.  The number that sticks out in my mind still is how the Bulls
actually did better without Rodman during his 3 years there than with
him and the guys replacing him were Kukoc, Bill Wennington, Dickey
Simpkins, Jud Buechler, Brian Williams -- guys who weren't as extreme
as Rodman, but did their jobs.  Rodman did yeoman's work, allowing
the other yeoman to do less, but that still doesn't make him a prince.



--- In APBR_analysis@y..., "mikel_ind" <msg_53@h...> wrote:
> Back in the early '90s, when Dennis Rodman decided to be a rebound
> specialist, I thought I noticed a subsequent decline in his
> teammates' rebounding numbers.  Laimbeer was actually in the
twilight
> of his career, so his rebounding rate might be expected to fade.
>
> Upon the Worm's move to San Antonio, David Robinson's career
peaked,
> in every measurable way except rebounding.  Same effect?  Not that
> it's surprising, but I wonder if overall team rebounds are helped
by
> a Rodman type; or if other categories are boosted at the same time;
> etc.
>
> I found 66 players who played significant minutes both with and
> without Rodman, in Detroit, San Antone, and Chicago.  I took a
simple
> ratio of their rebounding rate (with/without the worm), as well as
> other production rates.  Standardized, of course.
>
> Broken down by city:
>
> Tem Min. Pct. Sco. Reb. Ast. P.F. Stl. (TO) Blk. - Total
>
> Det 0.85 1.00 0.99 0.85 0.86 1.01 0.86 0.87 0.88 - 0.96
>
> SAn 0.93 1.01 1.05 0.88 0.99 0.97 1.06 0.95 1.01 - 1.00
>
> Chi 0.78 1.01 1.04 0.83 1.09 1.03 0.77 0.91 1.19 - 0.99
>
>
> Now by position:
>
> Pos Min. Pct. Sco. Reb. Ast. P.F. Stl. (TO) Blk. - Total
>
> C - 0.79 0.97 1.00 0.85 1.16 1.01 0.72 1.01 1.01 - 0.96
>
> F - 0.88 1.04 1.05 0.80 0.90 1.05 0.89 0.89 0.89 - 0.97
>
> G - 0.95 1.03 1.05 0.91 1.00 1.01 1.01 0.82 1.10 - 1.04
>
>
> Curiously, Rodman tended to cut into centers' minutes more than
into
> forwards'.  Then again, the Spurs' backup centers were J.R. Reid
and
> Antoine Carr, both really forwards.  Terry Cummings, too.  And the
> Bulls ran without centers a lot, when Rodman was there.
>
> Rodman's rising assist-enhancement rate seems to suggest he learned
> more nuances of offense, without being the actual ball-distributor?
>
> The data used in the ratios above were for whole player-seasons
with
> and without Rodman as a teammate.
>
>
> Mike G

--- In APBR_analysis@y..., bchaikin@a... wrote:
> because rodman played so many minutes and handled the ball so
little
> (identical touches per minutes for 92-93 in det and 93-94 in san),
and other
> power forwards on the 92-93 spurs handled the ball on offense more
than
> rodman did, in 93-94 other spurs had to handle the ball more in 93-
94 than
> they did in 92-93. that turned out to be robinson (either by design
or by
> natural flow of the game), who had 23% more ball possessions in 93-
94 versus
> 92-93 (1.40 possession factor in 93-94 versus 1.17 in 92-93), as
astounding
> change in the span of one season - i mean how many players can you
think of
> that were fulltime starters for four straight seasons and then all
of a
> sudden handled the ball on offense 23% more often (and not being a
point
> guard)? plus how often robinson shot, passed, got fouled, and
turned the ball
> over per ball possession was identical in both seasons - he
basically did not
> change his offensive game at all. yet because of the dramatic
increase in
> ball possessions due to having a player like rodman as a fulltime
floormate
> who just did not participate in the offensive scheme his scoring
average was
> by far a career best to that point...

I guess the surprising part is that Robinson stayed as efficient.  If
you take another player who touched the ball so little and put him on
the Spurs, Robinson also increases his touches.  But normally that
also means the defense can more easily defend Robinson, which I think
was true to some degree with Rodman, too.  In terms of efficiency,
Robinson was slightly increasing from 1990 to 1996, with '93 being a
down year actually.  He increased with Rodman's first year but not in
any way inconsistent with just maturing.  He was most efficient the
year Rodman left, though producing fewer points.  Since then, his
efficiency has been pretty constant, but lower than any of his first
several years.  His total points produced has come down steadily.

DeanO

Rodman was certainly an interesting player.  I wonder how much of an
asset he was to his teams.  I actually thought that Rodman on the
Pistons when he was able to score a little.  (Between 86 and 93
Rodman average about 8 to 9 ppg.  After 93, He never really scored
more than 5 ppg).  My feeling is that formulaic ratings of Rodman
tended to overrate him because he had so few shots missed and
turnovers when these stats were a function of his teammates often
playing 4 on 5 on offense and his being an offensive non-factor.  I
liked Rodman but thought he was most effective as a role player.

Dean noted that the Bulls were actually better in terms of w-l
without Rodman during their second 3-peat.  In fact, in the 97
playoffs Rodman was effectively displaced by Brian Williams to no
detriment.  (Rodman was fighting with the refs and would rack up
fouls too quickly to play.  Rodman was actually pretty valuable in
the 98 playoffs when Krause traded away the teams only other power
forward Caffey).

The Spurs also did not really miss him much either.  They traded him
away and replaced him with JR Reid and a broken down Charles Smith
and the team won 59 games (as opposed to the 58.5 wins they averaged
in Rodman's 2 years in SA).  He actually had breakdowns in the
playoffs both years on the Spurs.  The first year they lost to Utah
and he kind of freaked out and got frustrated and tried to injure
both Stockton and Malone.  (This was also when Madonna was sitting
courtside at all his playoff games).  The next year, Rodman got hurt
riding a motorcycle during the season and cameback for the playoffs
but got in a fight with his coach Bob Hill and often took of his
shoes on the bench and refused to join the huddles in the series
where Hakeem dominated DRobinson.

Of course, a lot of anti-Rodman stuff is not always fair.  His fights
with Bob Hill were a result of both him and Hill.  He always played
hard even though he was a little wierd.  However, We should recognize
that there were 2 Rodmans the role player on the bad boys who scored
a little more, rebounded a little less, and was an incredible
defender (I have no evidence of this other than perception).  The
later Rodman scored and defended less but rebounded like a fiend.
The first was an all-star quality player and the second was pretty
good but it did not seem that the teams he played on suffered much
from his absence.  It would been fun to see how much value he had as
a role player but by the time he reached that point of his career he
was too crazy to be an asset in a low-profile 15-20 mpg role (see his
Laker and Maverick tenures).

As for Rodman's stastical effect on his teams stars the best comp I
can come up with was Charles Oakley when Riley first got to the
knicks and kind of forbade him to shoot the ball.  Here are Oakley's
knicks stats pre-riley and with Riley when he refused to shoot as
compared with the Knicks primary scorer of that time Ewing:

           Oakley           Ewing

        ppg  rpg  fg%    ppg  rpg  fg%
88-89 12.9 10.5 51%     22.7 9.3  56.7%
89-90 14.6 11.9 52.4%   28.6 10.9 55.1%
90-91 11.2 12.1 51.6%   26.6 11.2 51.4%

with Riley
91-92 6.2  8.5 52.2%    24.0 11.2 52.2%
92-93 6.9  8.6 50.8%    24.2 12.1 50.3%

Don't really know what this might mean but its the best comparison I
could come up with.  Others I thought about were Dudley on NJ and his
effect on DC, Bo Outlaw on Orlando, and a few others but they seemed
too flawed to really compare.

It seems almost an unquestioned fact that Michael Jordan had a
positive effect on his teammates.  "He made his teammates better" is
the usual platitude.

He got them rings.  In many cases, he got them a job (Wennington,
Kerr, etc. etc.).  But were they measurably more effective in the
company of Mr. Jordan?

We know Cartwright had gaudy numbers in New York, which immediately
plunged upon coming to Chicago.  Same with Ron Harper coming from the
Clipps.

These guys were probably due for lean times, wherever they played.
But do all the players coming and going average out to something like
a trend?

I found 78 players who played significant minutes in consecutive
seasons with and without Jordan.  The field is definitely muddied by
the 1986 season and the 1995 season, when Jordan played only 15 or 20
games.  Frankly, I threw out some of the entrants.  And I'm not
considering the 2002 Wizards.

An average season with Jordan looks like this:

G Min Pct. Sco. Reb Ast Stl Blk - Total
59 20 .502 10.4 5.7 3.0 1.0 0.6 - 20.1

This is really weak.  It makes one remember how lame the Bulls were -
outside of Jordan - for several years.  Here's the summary of how
these 'standardized per-36-minute rates' with the Bulls compared to
the average rate with players' previous/subsequent team:

Min. Pct. Sco. Reb. Ast. Stl. Blk. - Total
0.88 0.98 0.96 0.95 1.15 1.73 1.11 - .98

Players tended to score and rebound less with Jordan around.
Shooting percents were also off a bit, and the total productivity was
suppressed, on average.

Here's a breakdown by position:

Centers:  Caldwell Jones, Corzine, Oldham, Waiters, Gilmore,
Cartwright, Sellers, Scott Williams, Stacey King, Wennington, Perdue,
James Edwards, Parish, Kleine, Longley, Simpkins.

Min Pct Sco Reb Ast Stl Blk - Total
18 .495 8.8 6.9 1.6 0.5 1.2 - 18.3
.93 .97 .95 .89 1.2 .67 1.01 - .95

Bulls' centers were notoriously lame, and quite a few in this bunch
were literally on their last legs.


Forwards:  Greenwood, Steve Johnson, Woolridge, Higgins, Sidney
Green, Gene Banks, Cureton, Oakley, Mike Brown, Charlie Davis, Nealy,
Levingston, Hopson, Horace Grant, Buechler, Blount, Rodman, Caffey,
Burrell, Pippen, Kukoc

Min Pct Sco. Reb Ast Stl Blk - Total
21 .515 11.6 7.6 2.3 0.9 0.7 - 22.1
.81 .98 .94 .98 1.09 .99 .95 - .97

Jordan didn't seem to enhance the performance of forwards, either.
Except in assists.


Guards:  Dailey, Whatley, Macy, Holton, John Paxson, Elston Turner,
Colter, Threatt, Sparrow, Sam Vincent, Hodges, Hansen, Tucker,
Darrell Walker, Ron Harper, Armstrong, Randy Brown, Kerr

Min Pct Sco. Reb Ast Stl Blk - Total
20 .493 10.2 3.0 4.7 1.3 0.2 - 19.3
.91 .98 .87 .92 1.02 .98 .67 - .95

It would now seem that the 'average' player did not see a statistical
boost from associating with Michael Jordan.  Yet Jordan, and the
Bulls, were not about average things.

Thanks to a very strong coach, the Bulls won 6 titles by discarding
the average-minded player and utilizing whatever excellence they
could glean from the role players.

When we look at minutes-distribution, we get a rather different view
of the Jordan Effect.

I have calculated productivity differences for the 78 player-seasons,
when players joined or left Jordan's Bulls.  In effect, I am figuring
the difference in production, in the minutes guys played with the
Bulls, compared to the same minutes with their previous (or
subsequent) team.


position pts reb ast stl blk
Centers +22 +3 +20 -3 +8
Forward +23 +1 +14 +15 +4
Guards- -5 -8 +8 +32 +1

This is an average production-per-season-per-player increase.  Here
are the composite totals:

po Pts Reb Ast. Stl Blk
C +445 +57 +404 -50 154
F +686 +41 +418 454 112
G -144 -213 226 904 +32

T 986 -115 1048 1307 298

This isn't a huge amount of anything, except maybe steals.  For 78
player-seasons, it's about 13 points per year, per player; 14
assists; 17 steals...

Once again, not sure what I have here.


Mike G

--- In APBR_analysis@y..., "mikel_ind" <msg_53@h...> wrote:
> It seems almost an unquestioned fact that Michael Jordan had a
> positive effect on his teammates.  "He made his teammates better"
is
> the usual platitude.
>

I haven't done the full study.  I did a quick look at offensive
efficiencies...

Players who definitely played more efficiently with Jordan and the
Bulls:  Ron Harper, Steve Kerr.

John Paxson didn't seem to be affected much.  B.J. Armstrong had his
best years with Chicago, but also a good one his first year away from
there.  Craig Hodges was about the same with or without.

Bill Wennington is a maybe.  Jason Caffey played more efficiently in
Chicago.  Luc Longley probably played better but he was also in his
peak years.  Will Perdue?  No big diff.  Stacey King played better
with Jordan.  Scott Williams played his first years in Chicago and
they were as good as some later years, so Jordan might have helped
him.  Hard to say with Horace Grant.

Rodman had his best offensive years before he decided to be weird,
which began well before he hooked up with Chicago.

It would be very nice if there weren't much complimentarity.  I think
there are cases where talents did compliment and guys fed off of
Jordan.  But really, it wasn't that many guys that did.  It basically
seems like those spot-up shooters get helped and a little of those
PFs, but that's not so clear.  Pippen's most efficient years were
definitely with MJ  -- he was less efficient in '94 and '95.  But he
was also not so efficient in '93 after good '91 and '92.  Maybe he
did need Jordan there, but also needed those couple years away from
him.  (Announcers did talk about his confidence problems before
Jordan left.)




> He got them rings.  In many cases, he got them a job (Wennington,
> Kerr, etc. etc.).  But were they measurably more effective in the
> company of Mr. Jordan?
>
> We know Cartwright had gaudy numbers in New York, which immediately
> plunged upon coming to Chicago.  Same with Ron Harper coming from
the
> Clipps.
>
> These guys were probably due for lean times, wherever they played.
> But do all the players coming and going average out to something
like
> a trend?
>
> I found 78 players who played significant minutes in consecutive
> seasons with and without Jordan.  The field is definitely muddied
by
> the 1986 season and the 1995 season, when Jordan played only 15 or
20
> games.  Frankly, I threw out some of the entrants.  And I'm not
> considering the 2002 Wizards.
>
> An average season with Jordan looks like this:
>
> G Min Pct. Sco. Reb Ast Stl Blk - Total
> 59 20 .502 10.4 5.7 3.0 1.0 0.6 - 20.1
>
> This is really weak.  It makes one remember how lame the Bulls
were -
> outside of Jordan - for several years.  Here's the summary of how
> these 'standardized per-36-minute rates' with the Bulls compared to
> the average rate with players' previous/subsequent team:
>
> Min. Pct. Sco. Reb. Ast. Stl. Blk. - Total
> 0.88 0.98 0.96 0.95 1.15 1.73 1.11 - .98
>
> Players tended to score and rebound less with Jordan around.
> Shooting percents were also off a bit, and the total productivity
was
> suppressed, on average.
>
> Here's a breakdown by position:
>
> Centers:  Caldwell Jones, Corzine, Oldham, Waiters, Gilmore,
> Cartwright, Sellers, Scott Williams, Stacey King, Wennington,
Perdue,
> James Edwards, Parish, Kleine, Longley, Simpkins.
>
> Min Pct Sco Reb Ast Stl Blk - Total
> 18 .495 8.8 6.9 1.6 0.5 1.2 - 18.3
> .93 .97 .95 .89 1.2 .67 1.01 - .95
>
> Bulls' centers were notoriously lame, and quite a few in this bunch
> were literally on their last legs.
>
>
> Forwards:  Greenwood, Steve Johnson, Woolridge, Higgins, Sidney
> Green, Gene Banks, Cureton, Oakley, Mike Brown, Charlie Davis,
Nealy,
> Levingston, Hopson, Horace Grant, Buechler, Blount, Rodman, Caffey,
> Burrell, Pippen, Kukoc
>
> Min Pct Sco. Reb Ast Stl Blk - Total
> 21 .515 11.6 7.6 2.3 0.9 0.7 - 22.1
> .81 .98 .94 .98 1.09 .99 .95 - .97
>
> Jordan didn't seem to enhance the performance of forwards, either.
> Except in assists.
>
>
> Guards:  Dailey, Whatley, Macy, Holton, John Paxson, Elston Turner,
> Colter, Threatt, Sparrow, Sam Vincent, Hodges, Hansen, Tucker,
> Darrell Walker, Ron Harper, Armstrong, Randy Brown, Kerr
>
> Min Pct Sco. Reb Ast Stl Blk - Total
> 20 .493 10.2 3.0 4.7 1.3 0.2 - 19.3
> .91 .98 .87 .92 1.02 .98 .67 - .95
>
> It would now seem that the 'average' player did not see a
statistical
> boost from associating with Michael Jordan.  Yet Jordan, and the
> Bulls, were not about average things.
>
> Thanks to a very strong coach, the Bulls won 6 titles by discarding
> the average-minded player and utilizing whatever excellence they
> could glean from the role players.
>
> When we look at minutes-distribution, we get a rather different
view
> of the Jordan Effect.
>
> I have calculated productivity differences for the 78 player-
seasons,
> when players joined or left Jordan's Bulls.  In effect, I am
figuring
> the difference in production, in the minutes guys played with the
> Bulls, compared to the same minutes with their previous (or
> subsequent) team.
>
>
> position pts reb ast stl blk
> Centers +22 +3 +20 -3 +8
> Forward +23 +1 +14 +15 +4
> Guards- -5 -8 +8 +32 +1
>
> This is an average production-per-season-per-player increase.  Here
> are the composite totals:
>
> po Pts Reb Ast. Stl Blk
> C +445 +57 +404 -50 154
> F +686 +41 +418 454 112
> G -144 -213 226 904 +32
>
> T 986 -115 1048 1307 298
>
> This isn't a huge amount of anything, except maybe steals.  For 78
> player-seasons, it's about 13 points per year, per player; 14
> assists; 17 steals...
>
> Once again, not sure what I have here.
>
>
> Mike G

--- In APBR_analysis@y..., "mikel_ind" <msg_53@h...> wrote:
> It seems almost an unquestioned fact that Michael Jordan had a
> positive effect on his teammates.  "He made his teammates better"
is
> the usual platitude.

Actually, upon reading a post in APBR about the old guys, I'm
wondering whether we can do this similar analysis on guys from
the '60's, like Russell, Pettit, Chamberlain.  Did Russell's
teammates play worse when he wasn't around?  Is there any indication
of this?  MikeG?


DeanO

--- In APBR_analysis@y..., "HoopStudies" <deano@r...> wrote:
>... I'm
> wondering whether we can do this similar analysis on guys from
> the '60's, like Russell, Pettit, Chamberlain.  Did Russell's
> teammates play worse when he wasn't around?  Is there any
indication
> of this?  MikeG?

Well, I'm not sure if there is major interest in the old-timers in
here, but I can certainly go thru the same techniques in any era.

Of course, away back then we didn't count turnovers, so "playing
efficiently" takes on a different meaning.  I tend to think of
productivity anyway, and the Big 3 Stats -- points, rebounds, and
assists -- are there, along with the vital Minutes Per Game.

Due to the stability of the Celtics during Russell's reign, there are
only some 23 guys who had consecutive seasons with and without Bill
Russell.  5 were carryovers from 1956 to '57, and 7 remained after
Russ retired in '69.  Only 11 came and went (to or from another team)
in the intervening 11 years.

Average productivity of these 23, in first or last season with
Russell:

Min Pct Sco. Reb Ast - Total
18 .439 11.0 5.2 2.0 - 19.1

And here is the fraction of productivity (>1.00 means more productive
with Russell):

Min Pct Sco. Reb Ast - Total
.85 .96 1.12 .91 .87 - 1.02

With minutes proportionately weighted, the projected increase in each
stat, for these 23 player-seasons:

Pts. Reb. Ast.
1407 -399 -156

It doesn't surprise me that Russ created more scoring for his
teammates, if only because he didn't take a lot of shots.  Neither is
it surprising he didn't leave many rebounds to be got.

The assist deficit is only about -7 per player-season -- not a lot,
but probably due again to the fact that most centers didn't do as
much of the passing.

I also don't know why guys shot lower percentages with Russell.

It would be nice to have a way to estimate Team Opponent Rebounds
before 1971.  Quite possibly, Russell's Celtics held a decided
rebound advantage over their opponents; in which case, all my
individual rebound ratings before 1971 are somewhat inaccurate.

I believe this issue has been broached before?


Mike Goodman

As a corollary to previous post, I'll look at Wilt's teammates with
and without him.

I expect to see guys getting fewer rebounds for sure; almost certainly
fewer points; perhaps more assists; and you might guess better
percentages (though it doesn't appear with Russell).

Some 76 players joined and/or parted company with the Dipper, playing
significant minutes in consecutive seasons with and without.

The year 1965 was spent half with the Warriors, and half with the
Sixers, so I omit that year, and '64 or '66 becomes the year of
measure.

1970 was a 12-game appearance in the regular season for Wilt, so it
is counted as a 'non-Wilt' year.

Breakdown by position:

Centers (12):

G Min Pct. Sco Reb Ast - Total
70 17 .458 9.8 7.7 1.8 - 19.4

Of course, no one really backed up Wilt, they moved to forward.
Johnny Kerr was the Sixers center when Wilt arrived, and he
languished on the bench or out of position, dealt at the end of '65,
so his numbers are omitted.

Centers' % change with Wilt:

Min Pct. Sco Reb Ast - Total
.69 1.01 .94 .85 .85 - .88

The big bite is in minutes, but across-the-board drop in
productivity.  Maybe the slight jump in shooting pct is just Wilt's
raising the standards.


Forwards (31):

G Min Pct. Sco. Reb Ast - Total
67 22 .464 11.2 5.7 2.1 - 19.3
0.85 0.99 0.92 0.93 1.08 - .90

Assists are up by the same ratio that scoring is down.


Guards (33):

G Min Pct. Sco. Reb Ast - Total
73 24 .463 10.8 3.1 3.8 - 18.0
1.02 1.00 0.97 0.90 0.93 - .94


And here are the projected production gains by these 76 player-
seasons:

Points
G Min Pct. Sco. Reb Ast - Total Pts. Reb. Ast.
66 21 .467 10.4 4.5 2.6 - 18.0 -118 -2215 405
0.89 1.00 0.95 0.87 0.97 - .91

Coincidentally the .91 overall productivity of Wilt's teammates is
almost the same reduction as he personally experienced in comparing
his regular-season to playoff rates.


Mike Goodman