A quick and dirty way to measure the predictiveness of a metric is to compare the performance of the first half of a season to the second half. - Seth Partnow, former head of analytics for the Milwaukee Bucks

Important background

  • The biggest advance in basketball over the past 20 years was the realization that 3 is more than 2.
  • The average distribution of shots has gradually shifted (and is still shifting!)
  • But what about NBA defense?

Past Performance, Future Performance

Of course this can all be explained

  • In the 24-25 season, the average defender was 14.5 feet away, meaning most shots are wide open!
  • The first half-second half split is totally arbitrary.
  • Defense simply looks different on the perimeter.

The key takeaway is that we are able to measure something that isn’t particularly meaningful BUT remains consistently reported.

Why I still can’t shake the figure

  • The NBA is changing: shot quality on two pointers has improved
  • It is difficult to leave data behind
  • Some teams do seem to improve (see: Celtics, Boston)