Analyzing My Perceptions About Basketball

The three or four of you who actually read my blog know that I played a Strat-o-Matic tournament where I created all-star teams for all four of the divisions in 2003-04, using Wins Produced as my benchmark.  This article is a review that, for the most part, takes a look at how I play Strat was influenced by my knowledge of Wins Produced.  In other words, I am reviewing myself.  At the bottom of the page, I will have a list of Wins Produced and Points over Par for all forty-eight players.

On thing I looked at was the correlation between a player’s Wins Produced Per 48 Minutes rate for the 2003-04 season and how many minutes that I played that player.  The correlation is a number between -1 and 1 which shows me how the variables move together (definition more or less borrowed from The Wages of Wins, the book I am reading which is where Wins Produced debuted), with positive numbers showing that they do move together, and negative numbers showing that they move away from each other.  The farther from zero they are, the stronger the correlation, positive or negative.  I am very disappointed with those results.  Separated by team, they are:

  • Atlantic: 0.39
  • Central: -0.23
  • Midwest: 0.50
  • Pacific: 0.15

I am going to add a couple of disclaimers.  For one, the Midwest team had Tim Duncan, Kevin Garnett, and Andrei Kirilenko.  If you look at their Strat-o-Matic cards, or if you know about basketball from that period, then you know that playing them was a no-brainer.  They combined to play 286 minutes, which is 38.3% of the team’s minutes.  On the other hand, there is a natural bias in play that is lowering these correlations-there were very few elite point guards that season compared to the other positions.  So, I also did correlations based on position.  These are still far from encouraging:

  • Center: 0.46
  • Power Forward: 0.67
  • Small Forward: 0.14
  • Shooting Guard: -0.32
  • Point Guard: -0.15

The data for Power Forwards and Centers was okay, but the other positions were clustered around 0 and largely negative.  Furthermore, players like Garnett, Duncan, Carlos Boozer, Lamar Odom, and Shaquille O’Neal were obvious choices to receive plenty of play time even if one does not know about Wins Produced, while Erick Dampier’s quality play showed me that he deserved more play time.  In addition, I already knew that Ben Wallace was one of the most productive players in the league last season according to Wins Produced.  It’s with the guards and the wings that this was less obvious, and if this tournament had been a test, I would have failed.

There are some specific examples that I can point to.  For one, Jerome Williams produced .243 Wins Per 48 Minutes in the 2003-04 season, which is outstanding.  However, his shooting columns are littered with replays.  As opposed to taking a shot, a replay is where the ball is passed around and another card is turned.  This scared me off, and Williams played a grand total of three minutes for the entire tournament.  Instead, I played Tayshaun Prince, Donyell Marshall, Keith van Horn, and Michael Redd  more minutes at small forward.  Of that list, only Marshall produced at twice the average level, and he played very poorly throughout the tournament, which prevented me from playing him more.

I used one other measurement to evaluate my decisions, using simple averages.  Using the players’ Wins Produced Per 48 Minutes rate for the 2003-04 season, I compared the number of wins a player would have produced if each player had played the same amount of time to the number of wins in the amount of time he actually played.  Again, I did not use the data from the tournament itself; I used the data from the season as a whole because a) each team only played three games in the tournament and b) the game is based on the statistics from the season itself.  I divided the number of wins based on actual minutes by those produced by equal minutes to come up with a percentage.  This way, I can see what percentage more wins I would have produced than I have I had just allocated minutes perfectly equally, and identifies my ability to evaluate the statistics on a Strat-o-Matic card because players produced wins at different rates.  Here are the percentages:

  • Atlantic: 101.84%
  • Central: 96.71%
  • Midwest: 102.72%
  • Pacific: 99.49%

Once again, I am disappointed by this data.  It implies that my evaluation of the talent ended up being about as effective as if I had just played each player for the same amount of time.  I would imagine that it would have been even worse if I had not already Wins Produced-based biases toward Ben Wallace, Shawn Marion, and Brent Barry.  (The first two played poorly, while Barry played exceptionally well in the tournament.  More on that later.)

Just for kicks, I decided to measure the correlation between Points Per 48 Minutes over the course of the season and minutes played in the tournament.  I must honestly say that I was not completely surprised judging by my previous findings, but it was nonetheless very disappointing.  I can’t say anything more.  Here they are:

  • Atlantic: 0.61
  • Central: 0.46
  • Midwest: 0.64
  • Pacific: 0.74

These three measurements lead me to conclude that even though I know about Wins Produced, it has not influenced my  evaluation of players using their Strat-o-Matic cards.  While Wins Produced Per 48 Minutes had very little correlation with minutes played in most cases, Points Per 48 Minutes had a large correlation.  I believe that both the Wins Produced correlations would be even lower, while the Points correlation would be higher, if I did not have some built-in biases toward certain players, such as Brent Barry and Ben Wallace.  If I had not had any bias, I can believe that this would have been the case.  I may say this once I crack open my 1997-98 Strat-o-Matic set.

I calculated in-tournament Wins Produced for each player.  I used defensive adjustments, unlike in my Olympic calculations.  Like my Olympic calculations, I used a standard average of the Adj48’s to use as my baseline position adjustments, as opposed to an average weighted using minutes played.  The reasoning is two-fold: a) I am never specifically told to use a weighted average, and b) I had difficulty when trying to find the weighted average in Microsoft Excel.  The position baseline Adj48’s are:

  • Point Guard: 0.525
  • Shooting Guard: 0.414
  • Small Forward: 0.337
  • Power Forward: 0.493
  • Center: 0.447

The table below has the Wins Produced for each player over the course of the tournament.  It also contains Points over Par Per 48 Minutes, which is a derivative of Wins Produced that shows how many points a player produces for his team, assuming that an average player produces zero points.  Thank you for reading, please comment, and please, please, please come back.

Name

Team

Position

Min

Adj48

WP48

WP

PoP/48

Barry, Brent

Pacific

SG

71

0.990

0.675

1.00

17.9

Dampier, Erick

Pacific

C

52

0.920

0.572

0.62

14.7

Posey, James

Midwest

SF

63

0.724

0.486

0.64

12.0

Kirilenko, Andrei

Midwest

SF

95

0.693

0.455

0.90

11.1

Brand, Elton

Pacific

PF

43

0.791

0.442

0.40

10.7

Garnett, Kevin

Midwest

PF

95

0.818

0.424

0.84

10.1

Bryant, Kobe

Pacific

SG

96

0.662

0.347

0.69

7.7

Boozer, Carlos

Central

PF

89

0.692

0.344

0.64

7.6

Billups, Chauncey

Central

PG

99

0.761

0.335

0.69

7.4

Bibby, Mike

Pacific

PG

79

0.747

0.321

0.53

6.9

Nash, Steve

Midwest

PG

59

0.716

0.290

0.36

5.9

Payton, Gary

Pacific

PG

65

0.699

0.273

0.37

5.4

Dunleavy, Mike

Pacific

SF

41

0.490

0.252

0.22

4.8

Duncan, Tim

Midwest

C

96

0.588

0.240

0.48

4.4

Odom, Lamar

Atlantic

PF

83

0.575

0.227

0.39

4.0

Mobley, Cuttino

Midwest

SG

58

0.539

0.223

0.27

3.9

Brown, P.J.

Central

C

37

0.566

0.218

0.17

3.7

Stojakovic, Peja

Pacific

PF

90

0.564

0.216

0.40

3.6

Hoiberg, Fred

Midwest

SG

32

0.529

0.214

0.14

3.6

O’Neal, Shaquille

Pacific

C

78

0.552

0.204

0.33

3.3

Ginobli, Manu

Midwest

SG

43

0.505

0.190

0.17

2.8

Foster, Jeff

Central

C

40

0.521

0.173

0.14

2.3

Miller, Brad

Pacific

C

41

0.518

0.170

0.14

2.2

Marion, Shawn

Pacific

SF

90

0.363

0.125

0.23

0.8

Miller, Andre

Midwest

PG

60

0.546

0.120

0.15

0.7

Wallace, Ben

Central

C

92

0.445

0.097

0.19

-0.1

Jefferson, Richard

Atlantic

SF

88

0.323

0.086

0.16

-0.4

Kidd, Jason

Atlantic

PG

91

0.506

0.080

0.15

-0.6

Camby, Marcus

Midwest

C

45

0.416

0.068

0.06

-1.0

McKie, Aaron

Atlantic

SF

41

0.291

0.053

0.05

-1.4

Redd, Michael

Central

SG

92

0.353

0.037

0.07

-1.9

Miller, Reggie

Central

SG

56

0.346

0.031

0.04

-2.1

Kittles, Kerry

Atlantic

SG

43

0.344

0.029

0.03

-2.2

Cassell, Sam

Midwest

PG

50

0.453

0.027

0.03

-2.3

Prince, Tayshaun

Central

SF

74

0.242

0.004

0.01

-3.0

Van Horn, Keith

Central

PF

36

0.333

-0.016

-0.01

-3.6

Brown, Kwame

Atlantic

C

50

0.327

-0.021

-0.02

-3.7

Marshall, Donyell

Central

SF

58

0.145

-0.093

-0.11

-6.0

Marbury, Stephon

Atlantic

PG

54

0.290

-0.136

-0.15

-7.3

Williams, Jerome

Central

SF

3

0.102

-0.136

-0.01

-7.3

McGrady, Tracy

Atlantic

SG

91

0.136

-0.179

-0.34

-8.6

Jamison, Antwan

Midwest

PF

47

0.114

-0.234

-0.23

-10.4

Jones, Eddie

Atlantic

SF

38

-0.003

-0.241

-0.19

-10.6

Thomas, Kenny

Atlantic

PF

37

0.058

-0.291

-0.22

-12.1

Dalembert, Samuel

Atlantic

C

52

0.038

-0.310

-0.34

-12.7

Blount, Mark

Atlantic

C

51

0.028

-0.320

-0.34

-13.0

Sura, Bob

Central

PG

38

0.006

-0.420

-0.33

-16.1

Christie, Doug

Pacific

SG

3

-0.261

-0.576

-0.04

-21.0

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s