Basketball Stats

Executive Summary We tried to find out how certain basketball statistics affect winning percentage for a NCAA Division I basketball team. We used the entire NCAA division I 1999-2000 season statistics. We considered the following statistics: Field Goals, Free Throws, Personal Fouls, Turnovers, 3 Pointers, Blocks, & Steals. Our conclusion is that while a rise in each stat had some affect in the rise or fall of winning percentage, we could not determine a single stat that had a direct affect on the dependent variable (Winning Percentage). Our results were more effective when we ran the test on how the combination of all stats affected winning percentage, however, this would be obvious given the nature of our study. While this study did not produce the result we wanted, we believe that we could use the information learned from this study and develop a study that would be more effective. Introduction Coaches are always looking for a better understanding of what makes up a winning team. This knowledge would help them in recruiting athletes that could improve the team’s statistics in the areas we observed. We took the entire statistical breakdown from the 1999-2000 season and were hoping to find any key statistical areas that could be directly related to winning percentage. Methodology We will run regression analysis on how each independent variable affects the dependent variable individually and in total. The dependent variable is winning percentage, and the stats we measured were selected as the independent variables. The independent variables that we chose were: Field Goals per Game, Free Throws per Game, Assists per Game, Fouls per Game, Turnovers per Game, 3 Pointers per Game, Blocks per Game, Steals per Game & Rebounds per Game. Our statistics were gathered from www.cbssportsline.com. We will use these statistics to run a regression analysis to see if any one statistic could be used to predict winning percentage. Findings This section includes the actual statistical calculations. It shows the calculations of how each statistical variable affected winning percentage individually and how in combinations the same statistical variables affected winning percentage. The data we gathered from our analysis are presented in a formal way on the following page. Regression Steals Turnovers Committed Made Field Goals Made Free Throws Statistics Per Game Per Game Per Game Per Game Multiple R 0.228547 0.429836 0.510207 0.384329 R Square 0.052234 0.184759 0.260311 0.147709 Adjusted R Square 0.049234 0.182179 0.257970 0.145011 Standard Error 0.177928 0.165020 0.157188 0.168728 Observations 318 318 318 318 Regression Made 3 Pointers Blocked Shots Personal Fouls All Variables Statistics Per Game Per Game Per Game Per Game Multiple R 0.147094 0.366930 0.276749 0.800440 R Square 0.021637 0.134637 0.076590 0.640704 Adjusted R Square 0.018541 0.131899 0.073668 0.632591 Standard Error 0.180778 0.170017 0.175627 0.110607 Observations 318 318 318 318 The findings of listed in these tables show us how a jump in each independent variable considered individually would affect the dependent variable. The last statistical category listed show the relationship observed when all independent variable are considered together. Each independent variable considered separately had very low multiple R values which means that they could not significantly be used to predict the dependent variable of winning percentage. As you can see this last study gives us a better Multiple R value and a better R Squared value as well. We ran the multiple regression analysis eliminating variables that had lower R values but found no improvement from our original findings. Conclusions and Recommendations We were trying to determine what statistic or group of statistics could be used the help predict a teams winning percentage. Our study does not accomplish what we started out wanting to find, however, it does provide us with some direction with which to formulate future studies. The study does suggest that the independent variables considered as a whole could be used to predict winning percentage but this is an obvious result. You would expect this result given the nature of the variables we studied. It is our opinion that further studies would have to be conducted to achieve the desired results we were looking for.