Post by eric on Dec 28, 2015 18:21:18 GMT -6
The four factors are...
eFG%
turnovers
rebounding
free throw rate
With these and only these you can totally describe a team's offense and defense, and therefore their expected win-loss record. But how do we weight them? Dean Oliver, their inventor, suggests 40/25/20/15. In this paper I will demonstrate that everything that guy just said is b***shit.
I calculated the relative Rtgs and factors for every team-season from 1984 to 2013. I then looked for couplets where three of the four factors agreed to within .001, and graphed the the difference in Rtg+ against the difference in fourth factor. For example, here is the graph for own eFG%:
I did this with all eight factors and obtained a model of 65/20/10/05, then tried the model on the four factor data for the 2014 season:
Clearly this method is awesome and I am right. This is my conclusion.
A+
eFG%
turnovers
rebounding
free throw rate
With these and only these you can totally describe a team's offense and defense, and therefore their expected win-loss record. But how do we weight them? Dean Oliver, their inventor, suggests 40/25/20/15. In this paper I will demonstrate that everything that guy just said is b***shit.
I calculated the relative Rtgs and factors for every team-season from 1984 to 2013. I then looked for couplets where three of the four factors agreed to within .001, and graphed the the difference in Rtg+ against the difference in fourth factor. For example, here is the graph for own eFG%:
I did this with all eight factors and obtained a model of 65/20/10/05, then tried the model on the four factor data for the 2014 season:
Clearly this method is awesome and I am right. This is my conclusion.
A+
