Slide 1Analysis MethodSlide 3Physical AttributesSlide 5Statistical Performance“Defense Wins Championships”All-Star StatusAll Star/Non All Star DifferenceFranchise PlayerLarry Bird RuleLinear RegressionTwo-Factor AnovaCorrelation Test for Team/YearsWithout ReplicationDifference by Larry Bird RuleFinal RegressionConclusionsThank YouNBA Statistical AnalysisEcon 240AGroup Project II12/3/08Jeff LeeZhen TianTsung-Ching HuangOystein TennumChristian TreubigEddie BedachAnalysis Method Statistics CollectedSalary – Dependent VariableHeightWeightPPG,RPG,APG,SPG,BPGAll-Star StatusNumber of teams played forAnalysis Method 300 player database created50 players from each division - Starters and Second string from each team Exploratory Data AnalysisInvestigation of various factors on salaryOrdinary Least Square - T-statistic, F-statisticAnalysis of VariancePhysical AttributesHeight and weight found to be insignificant.Tall/Weak is undesirableShort/Strong is undesirableDependent Variable: SALARYMethod: Least SquaresSample: 1 300Included observations: 300Variable Coefficient Std. Error t-Statistic Prob. HEIGHT 151029.1 103165.2 1.463954 0.1443WEIGHT 16999.88 14633.30 1.161726 0.2463C -9864304. 6208414. -1.588861 0.1132R-squared 0.041654 Mean dependent var 5911743.Adjusted R-squared 0.035223 S.D. dependent var 5018847.S.E. of regression 4929667. Akaike info criterion 33.66936Sum squared resid 7.24E+15 Schwarz criterion 33.70631Log likelihood -5064.238 F-statistic 6.476267Durbin-Watson stat 1.925366 Prob(F-statistic) 0.001765Why I was picked after you?U’re shorter yet fatter…Physical AttributesCreated dummy variable: - Body = height * weightBody was found to be significantDependent Variable: SALARYMethod: Least SquaresSample: 1 300Included observations: 300Variable Coefficient Std. Error t-Statistic Prob. BODY 338.0742 94.68063 3.570680 0.0004C -101890.1 1707911. -0.059658 0.9525R-squared 0.040897 Mean dependent var 5911743.Adjusted R-squared 0.037690 S.D. dependent var 5018847.S.E. of regression 4923359. Akaike info criterion 33.66350Sum squared resid 7.25E+15 Schwarz criterion 33.68813Log likelihood -5064.357 F-statistic 12.74976Durbin-Watson stat 1.933402 Prob(F-statistic) 0.000415Statistical Performance Most Important Factors1. Blocks2. Assists3. Points4. Rebounds5. StealsDependent Variable: SALARYMethod: Least SquaresSample: 1 300Included observations: 300Variable Coefficient Std. Error t-Statistic Prob. APG 541720.3 159737.9 3.391307 0.0008BPG 1727532. 488891.3 3.533571 0.0005PPG 502778.7 52958.75 9.493781 0.0000RPG 424591.4 130701.2 3.248565 0.0013SPG -549798.2 664644.0 -0.827207 0.4088C -2748804. 465895.7 -5.900042 0.0000R-squared 0.641952 Mean dependent var 5911743.Adjusted R-squared 0.635883 S.D. dependent var 5018847.S.E. of regression 3028477. Akaike info criterion 32.70475Sum squared resid 2.71E+15 Schwarz criterion 32.77865Log likelihood -4916.065 F-statistic 105.7824Durbin-Watson stat 1.682747 Prob(F-statistic) 0.000000“Defense Wins Championships”Dependent Variable: SALARYMethod: Least SquaresSample: 1 300Included observations: 300Variable Coefficient Std. Error t-Statistic Prob. DEFENSE 511695.3 43166.52 11.85399 0.0000OFFENSE 78918.61 6159.982 12.81150 0.0000C 2453181. 266479.8 9.205881 0.0000R-squared 0.553566 Mean dependent var 5911743.Adjusted R-squared 0.550570 S.D. dependent var 5018847.S.E. of regression 3364611. Akaike info criterion 32.90544Sum squared resid 3.37E+15 Schwarz criterion 32.94239Log likelihood -4949.269 Hannan-Quinn criter. 32.92022F-statistic 184.7563 Durbin-Watson stat 1.953691Prob(F-statistic) 0.000000To investigate this, two dummy variables were createdOFFENSE = PPG*APGDEFENSE = RPG*BPG*SPGDefense was found to be much more significant. Teams are willing to pay more for defensive starsAll-Star StatusHighly SignificantAll Stars make over an average of $3,426,702 more than non all star playersDependent Variable: SALARYMethod: Least SquaresSample: 1 300Included observations: 300Variable Coefficient Std. Error t-Statistic Prob. ALLSTAR 3426702. 624127.2 5.490390 0.0000APG 462621.4 153067.2 3.022343 0.0027BPG 1223729. 475336.2 2.574450 0.0105PPG 363983.5 56493.55 6.442921 0.0000RPG 425565.1 124687.2 3.413062 0.0007SPG -569773.6 634071.1 -0.898596 0.3696C -1448944. 503581.2 -2.877281 0.0043R-squared 0.675249 Mean dependent var 5911743.Adjusted R-squared 0.668622 S.D. dependent var 5018847.S.E. of regression 2889123. Akaike info criterion 32.61378Sum squared resid 2.45E+15 Schwarz criterion 32.70000Log likelihood -4901.375 F-statistic 101.8850Durbin-Watson stat 1.727203 Prob(F-statistic) 0.000000All Star/Non All Star Difference5 Yrs for $50,000,000 6 Yrs for $118,000,00016.1 points/game 16.8 points/game5.1 rebounds/game 5.8 rebounds/game2.2 assists/game 1.8 assists/game9 Years for 3 Teams 10 Years for 2 Team78 inches + 225 lbs 82 inches + 230 lbsAll-Star? No All-Star? Yes, only 2005Why I earn much less than u?Franchise PlayerA player who plays for one team for the majority of his career is considered a “Franchise Player.”Create a dummy variable ‘FRANCHISEVET.’Product of Variables Franchise and VeteranFranchise = 1 if player has played for one team onlyVeteran = 1 if Years Played >5Franchise Players make nearly 3 million more per year, on averageDependent Variable: SALARYMethod: Least SquaresSample: 1 300Included observations: 300Variable Coefficient Std. Error t-Statistic Prob. FRANCHISEVET 2997136. 774263.4 3.870952 0.0001APG 554151.9 156114.3 3.549654 0.0004BPG 1463554. 482542.9 3.033003 0.0026PPG 471283.8 52382.18 8.997027 0.0000RPG 432064.7 127723.9 3.382803 0.0008SPG -411064.2 650417.5 -0.632001 0.5279C -2616565. 456510.7 -5.731662 0.0000R-squared 0.659316 Mean dependent var 5911743.Adjusted R-squared 0.652363 S.D. dependent var 5018847.S.E. of regression 2959150. Akaike info criterion 32.66168Sum squared resid 2.57E+15 Schwarz criterion 32.74790Log likelihood -4908.583 F-statistic 94.82814Durbin-Watson stat 1.604668 Prob(F-statistic) 0.000000Larry Bird RuleExceed the salary cap to re-sign their free agentsUp to maximum salaryUp to 6
View Full Document