Breaking down PER

[Keller61]

uPER = (1 / MP) *
[ 3P
+ (2/3) * AST
+ (2 - factor * (team_AST / team_FG)) * FG
+ (FT *0.5 * (1 + (1 - (team_AST / team_FG)) + (2/3) * (team_AST / team_FG)))
- VOP * TOV
- VOP * DRB% * (FGA - FG)
- VOP * 0.44 * (0.44 + (0.56 * DRB%)) * (FTA - FT)
+ VOP * (1 - DRB%) * (TRB - ORB)
+ VOP * DRB% * ORB
+ VOP * STL
+ VOP * DRB% * BLK
- PF * ((lg_FT / lg_PF) - 0.44 * (lg_FTA / lg_PF) * VOP) ]

factor = (2 / 3) - (0.5 * (lg_AST / lg_FG)) / (2 * (lg_FG / lg_FT))
VOP = lg_PTS / (lg_FGA - lg_ORB + lg_TOV + 0.44 * lg_FTA)
DRB% = (lg_TRB - lg_ORB) / lg_TRB

I have questions about the merit of this stat.

- Who decided that an assist is worth 2/3 of a point? That seems rather unscientific.

- Why are rebounds worth more than assists? I tend to value assists more.

- Why are steals worth more than blocks? A successful steal might generally be more valuable than a successful block, but block numbers tend to indicate how well a player contests shots. A near-block can result in an altered shot and force a miss. A near-steal however, may be from gambling and could give the opponent an easy opportunity.

- Why is league average rebounding % used instead of team average? Some players affect their team's ORB% by the shots that they take, and this can even be a strategy - putting up tough shots after drawing in defenders, or taking shots when they see that teammates have good position for rebounds.

- Why is league average VOP (points per possession?) used instead of team VOP?

- Why are STLs and OREBs multiplied by average PPP, when they usually result in better-than-usual scoring opportunities?

- Why are fouls included as a negative? Fouls are often good plays to prevent easy points.

My biggest criticism of this stat, however, is the idea that it can be universally applied to every player in every situation. Although I suppose you could say that about most stats.

I don't understand why this stat has become so widely used. It is merely John Hollinger's opinion of the relative worth of points, rebounds, assists, etc. I find it useful for vaguely categorizing players; for example, if I want to make a list of the best players in a given year I usually start by sorting them by descending PER just to get a vague idea. However, in terms of side-by-side comparisons, I don't think it has much value. In order to really compare two players, you have to be able to break down their stats, compare situations, and analyze the stats in context. PER gives you nothing in terms of context.

[Ice Man]

Answer to all question before the final one: Because.
Answer to the final one: Because somebody with authority created an all-in stat

If you or I did the task, we would do it differently. All our results would be highly correlated though. Kevin Durant and LBJ would be at the top, David West would be in the middle, and Marquis Teague would be at the bottom.

I like PER for following a player's career -- seeing the rise and fall. Most of the problems of the arbitrariness fall away with that usage because the player pretty much always is good and bad at the same things, so the change in PER measures just how good or just how bad. It's great for that snapshot, much easier than looking at, say, 5 variables and then trying to juggle in one's head, hmmm fewer shots but efficiency hanging in there, rebounds down but assists slightly up, fewer blocks, etc.

[Dr Positivity]

I've always thought that Hollinger made the stat by taking the preconception of how good every player is, then roughly adjusting the variables to whatever combination best fit these preconceptions