Saturday, January 12, 2019

Calculating defensive rating for guarding opponents

In 2011, this post broke down the defensive calculations for the Value Add Basketball rankings for all college players from 2002 to the present. While the rankings received national praise, this continues to be the most underappreciated aspect of the rankings. We will periodically list the four defensive considerations in the Notes field to show each qualifying players grade from 0 to 10 in each:

DefR = Defensive Rebounding

Blk = Blocked Shots

Stl = Steals

Guard = The most important defensive factor is the ability to guard opponents to deny them the ball, force them into bad shots, and create turnovers, and a small consideration for avoiding fouling, all weighted by the level of offenses faced and the minutes on the floor on defense.

For example, Zion Williamson had notes field on January 19, 2019 that read DefR 9^ Blk 10^ Stl 10^ Guard 7 for his rating of 0 to 10 in each of the four categories. The math that went into calculating that he was the top defender in the country and taking away 3.52 points per game from an opponent compared to how many they would score against a replacement defender. However, the four ratings on the 1 to 10 scale at least gives an idea of which aspects of defense are strong or weak.

While other systems rank the first three, they ignore the fact that these three together are less important than guarding your opponent and a player who scores a lot of points but allows opposing players to cut to the rim in 100-90 losses is not nearly as valuable as a player who scores even half as many points and helps his team pull out 50-40 wins.

The following is a breakdown of how these factors play out in a given game, in this case Marquette's defensive effort at Creighton in 2019, which was not good overall but was overcome by Markus Howard setting the Big East record with 53 points that night.

Running total
More defensive rebounds than one-third of missed shot, worth one point each
(MU grabbed 29 of 35 CU misses, 4.85 more than typical 69%)
more blocks than typical based on number of 2-point shots taken worth
0.67 points each (CU 32 two-point shots typically result in 3.52 blocks,
but MU blocked 6)
more steals than expected based on trips down the court worth one point
each (MU 80 defensive possessions typical 5.68 steals, but only had 2 steals)


Total Points taken away (or added if a negative) based on blocks, steals and

actual points scored by opponent

anticipated points allowed if if blocks, steals and defensive rebounds average
(total of the previous 2 numbers to reverse engineer)
how many points would have been scored if 1.027 per possessions in the game

how good was the guarding defense based on points less than expected
not counting
blocks, steals and rebounds
how good or bad (if negative) the guarding defense was after using the
number of minutes played in the game (200 if ends in regulation, add 25
for each overtime)
based on the players minutes played, how many points did the player take
away or add with good or bad defense
That was the defensive side to this post on the Marquette CBS 247 page. Beyond those defensive
considerations, the following are the only possible plays in a game that Markus Howard did
NOT have during the game and how valuable each of those plays is.

Only things Markus Howard did NOT do in game
Made 2-point show and drew foul (unassisted)
Made 2-point show and drew foul (assisted)
Draw 3-shot foul
Offensive Rebound
Draw 1-and-1 foul shots
Foul drawn but no free throws

Missed deadball FT, no rebound chance
Miss front end of 1-and-1

It appears the one thing a player can do that hurts their team the most in all of basketball is missing the front end of a 1-and-1, since an average free throw shooter scores 1.11 points per one-on-one opportunity, so in the just under 3 out of 9 times the front end is missed it means the player secured 0.00 or the average 1.11 points. However, because the team does have a one-third chance of an offensive rebound, there is a chance the get the miss and score so it counts as a -0.74. We do not have a way to break down the slightly lower chance of an offensive rebound on a free throw miss verses and attempted field goal.

No comments:

Post a Comment