# An Introduction to Four Factors Rating

## A rating system based on the four factors of basketball success

Four Factors Rating (FFR) is a model that rates teams based on how well they fare in the four factors of basketball success, pioneered by Dean Oliver.

The model is two linear regressions, one for offense and the other for defense. The current iteration of FFR is a 5 year sample (2013–17), and the inputs are the standard deviations above/below average for the given season for each of the offensive and defensive four factors.

The first figure shows the regressions statistics for Offensive FFR, and as shown by the coefficient, the model weighs scoring efficiency from the floor very highly. This is in line with the conventional thinking that eFG% is the most important of the four factors. A team’s ability to get to the free throw line is about a quarter of the impact as efficiency from the field while limiting turnovers and crashing the offensive glass are close to equal significance.

Defensively the model doesn’t give quite as much credit to teams for defending from the field with the coefficient only about 72% as the offensive eFG%. The other significant difference in the offensive and defensive models is the value of rebounding.

The value of defensive rebounding in the model is being reduced, and that reflects the current trends in basketball that are leading to success. The only variable that has more value on the defensive end is TOV%, which in conjunction with the DRB% coefficient points to transition offense (something not specifically being measured in this model currently) becoming increasingly important. Teams may have to gamble on the defensive boards to get consistent transition looks off of missed shots, but the teams able to successfully end defensive possessions with one less man are able to see immediate reward in the form of easy transition shots.

For win projections, +1 FFR is worth 2.5 wins above average.

*Four Factors Rating for the 2017–18 will be at least partially projection-based until sometime in December, and data to build models is courtesy of Basketball-reference.com*