APBRmetrics

A bit of analysis of PER vs PER2.0 for 2016-2017 data:

On average guys 20 or higher on PER are 21.4% higher on PER2.0 for an average gain of 5 points.
Guys 15-20 on PER only gain 6.7% or a bit more than 1 point on average.
Guys 10-15 on PER lose an average of 8.4% or about 1 point.
Guys below 10 on PER lose 74% on average or about 5 points.

So PER2.0 compared to PER changes things dramatically on the tails but every little in the middle.

Would be interesting to see by position, by usage, etc. Justin, do you plan to do any of that in a follow-up?

The above was not minutes weighted. Might affect the results. May check later.

Crow wrote:A bit of analysis of PER vs PER2.0 for 2016-2017 data:

On average guys 20 or higher on PER are 21.4% higher on PER2.0 for an average gain of 5 points.
Guys 15-20 on PER only gain 6.7% or a bit more than 1 point on average.
Guys 10-15 on PER lose an average of 8.4% or about 1 point.
Guys below 10 on PER lose 74% on average or about 5 points.

So PER2.0 compared to PER changes things dramatically on the tails but every little in the middle.

Would be interesting to see by position, by usage, etc. Justin, do you plan to do any of that in a follow-up?

Well PER was largely controlled by its assisted factor. Since box score stats have no information on how many shots were assisted or unassisted, everyone on the same team got the same adjustment. Once you change that, it completely transforms and widens the ratings.

I'm not sure what I want to do for a follow-up, if anything. I might wanna see if it's more useful or maybe do a team adjustment and some way of factoring in shot defense.