The Process
To begin finding the value over expected role, the roles must first be defined. In order to get the roles distance similarity was chosen as the method. Each players defending frequency across the court was taken and converted to a percentile score. For example, Clint Capela defends 41% of his shots at less then 5 FT from the rim putting him in the 77th percentile of the league for rim defense frequency.
Each role was based on purely where you defend. Paint, midrange (short and long), 3PT, and combinations of those three. Each player was then compared against all 3 through distance similarity to find their closest role.
With each of the role scores for every player, and their FG% defended in each of these area, we can calculate the average FG% defended for each area for each role. This number allows us to evaluate each player against their role and find value over expected. A big man would not be held to high standards on 3PT shooting, but will be held to higher standards on rim defense.
Since roles aren't static each player is taken as a combination of roles and weighed against that mix to find the residual against league avg. The residual is FG% defended over (or under) expected FG%.
These roles do not include any disruption statistics, such as blocks, steals, or deflections. To account for this, the same process was repeated with these stats to find a players block, steals, and deflection over role expectations.
In order to weigh these stats against volume, they are converted to points saved. This makes it so role players, who defend elitely in extremely low volumes do not show up as the best defenders in the world, compared to someone like Victor Wembanyama who faces multiple shots from every area in the court. This is done by taken the residual for each player in each area and multiplied by the number of shot attempts and their worth (2p & 3p).
The same thing is done for the disruption factor, where steals are given 1 point because one possession saved which on average is worth 1 point, blocks are worth .7 points since each block doesn't always result in a saved possession (out of bounds), and deflections are worth .3 points since they are more of a momentum shifter.
Click any column header to sort. Low GP players are highlighted in amber — small samples produce noisy estimates.
Loading...
Limitations
Doesn't take into account shot deterrence — players like Wembanyama deter shots from being taken.
Players on a good defensive team benefit since great teammates inflate their shot suppression stats.
Strengths
Role Adjusted — Takes into account where players are expected to have good defense.
Volume weighted & two dimensional — you can see where a player gets their value from (either disruption or shot quality).