In the NBA in 2018–19, the 2P% within three feet of the basket was 66%.
From three feet up to the 3P line, 2P% was pretty consistent at 40%.
The league-wide 3P% was 36%. The eFG% of 3PA was 53%.
So, three-pointers had an eFG% that was 13 percentage points higher the eFG% of the 2PA outside three feet.
Given this, the analytics say that teams should shoot nothing but close-range 2PA and 3PA, don’t they? In other words, that teams shouldn’t shoot any mid-range 2PA at all, right?
Wrong. That’s an invalid conclusion. There’s nothing analytical about it.
Rather, the conclusion is positivistic. It’s based on the belief that there are absolute laws in human sciences, like there are in natural sciences.
I’m not arguing against shooting a lot of 3PA. It can be a good idea. Yet the strict rule of “no mid-range 2PA” is non-optimal for at least two reasons.
First, the rule makes the offense fragile and predictable. The defenses can adjust because they don’t have to worry about anyone shooting from the mid-range. They can for example use the drop coverage in pick-and-rolls.
Secondly, the “no mid-range 2PA” rule leads to a non-optimal shot selection. On the average, 3PA do have a higher eFG% than mid-range 2PA. But in games the players are not selecting between these general categories but rather between singular shot opportunities that arise in a possession.
And some mid-range 2P opportunities have a higher expected eFG% than some 3P opportunities . So, if a team is taking no mid-range 2PA, it doesn’t imply that the team is playing smartly. Rather, it implies that the players are probably trading better 2P opportunities for worse 3PA.