The Four Factors Team Ratings was created as an alternate to most rating systems which rely on point differential, and not the actual performance of the team. I gathered data from every NBA team since 1974 (and the ABA from 1974-76), ran it through a linear regression, and the resulting model rates every team compared to league average for that season. This is a good stat to see just how much a certain team dominated, or was dominated by, its peers. The term "Four Factors" comes from Dean Oliver and the four primary stats that explained most, if not all, of team performance. Possessions can only end so many ways, and this model accounts for all of those ways. All data is from http://www.basketball-reference.com
Possessions can either end in a shot, foul, or turnover. The first part of the stat I'll talk about is the turnovers, since that ends the possession. Both a teams turnover % (TOV%) and the opponents turnover % (O. TOV%) are included in the 8 input variables for the linear regression. When a team takes a shot, the efficiency of said shot can be shown through effective field goal % (eFG%), and the efficiency of opponent shots can be shown through opponents effective field goal % (O. eFG%), and both of these are inputs variables for each team. When a shot is missed, the chance one team or another grabs the rebound can be expressed with defensive rebounding % (DREB%) and offensive rebounding (OREB%), which are both input variables. The last two variables are to account for fouls being drawn, and to show that I will use the ratio of free throws to field goal attempts for both team (FT/FGA) and opponent (O. FT/FGA). That output variable for the samples will be each team's Offensive Rating - Defensive Rating, their Net Rating. Teams will be compared to league average for their respective season in the formula with the same variables proceeded by LG. example: (LG.eFG%) is league average effective field goal%.
The model for the equation prior to 2016 is (rounded for ease of reading) is:
Four Factors Rating =.004+1.47(eFG%-LG.eFG%)+1.52(LG.TOV%-TOV%)+.43(OREB%-LG.OREB%)+.32(FT/FGA-LG.FT/FGA)+1.42(LG.eFG%-O.eFG%)+1.4(O.TOV%-LGTOV%)+.44(DREB%-LG.DREB%)+.3(LG.FT/FGA%-O.FT/FGA)
For the 2016 season, Four Factors Rating will be separated by offense and defense, and in the 2017 season the standard deviation for input variables will be accounted for.
Possessions can either end in a shot, foul, or turnover. The first part of the stat I'll talk about is the turnovers, since that ends the possession. Both a teams turnover % (TOV%) and the opponents turnover % (O. TOV%) are included in the 8 input variables for the linear regression. When a team takes a shot, the efficiency of said shot can be shown through effective field goal % (eFG%), and the efficiency of opponent shots can be shown through opponents effective field goal % (O. eFG%), and both of these are inputs variables for each team. When a shot is missed, the chance one team or another grabs the rebound can be expressed with defensive rebounding % (DREB%) and offensive rebounding (OREB%), which are both input variables. The last two variables are to account for fouls being drawn, and to show that I will use the ratio of free throws to field goal attempts for both team (FT/FGA) and opponent (O. FT/FGA). That output variable for the samples will be each team's Offensive Rating - Defensive Rating, their Net Rating. Teams will be compared to league average for their respective season in the formula with the same variables proceeded by LG. example: (LG.eFG%) is league average effective field goal%.
The model for the equation prior to 2016 is (rounded for ease of reading) is:
Four Factors Rating =.004+1.47(eFG%-LG.eFG%)+1.52(LG.TOV%-TOV%)+.43(OREB%-LG.OREB%)+.32(FT/FGA-LG.FT/FGA)+1.42(LG.eFG%-O.eFG%)+1.4(O.TOV%-LGTOV%)+.44(DREB%-LG.DREB%)+.3(LG.FT/FGA%-O.FT/FGA)
For the 2016 season, Four Factors Rating will be separated by offense and defense, and in the 2017 season the standard deviation for input variables will be accounted for.