When in doubt, simulate
by Ken Pomeroy on Friday, February 24, 2012
I’ve been concerned about the sample size of this week’s work regarding 3P%. We’re only looking at 8-10 games in each half of the conference season. As Kevin Pelton gently reminded me, there are fewer three-point shots than two-point shots and thus there will be more noise in the 3P%. Still, if there is so much noise over 8-10 games that you can’t see a signal, just think about how much noise there is over six games. Consider that when your bracket goes up in flames.
Still, the offensive 3P% scatterplot shown yesterday was troubling. If the defense has no influence on 3P%, why does offensive 3P% show no relationship between the first and second half of conference play? The only solution is to simulate the conference season and see what the three-point percentage scatterplot would look like if team 3P% were truly independent from shot to shot.
But first,…