OneBadLT123

I see PER used a lot, and I understand the basics somewhat, but how exactly is PER calculated? And what does it show/explain in terms of a players game?

I am just trying to fully understand this way of showing a players overall efficiency.

__________________
-
OneBadLT123's favorite 07-08 Rockets Videos:
41 Years of Rockets Tradition
Game # 22 in a Row Won
Saluting the Rockets
OneBadLT123's All-time Favorite Rockets Videos:
The Best Hakeem Mix
Hakeem Owning Robinson 95
Hakeem Owning Robinson Again
Post Game 95 Finals
Johnson At The Buzzer!
Tmac 13 in 33
Random Vids:
NBA on NBC Memories
Vince Carter 2000 Slam Dunk Contest

Diophantos

You can read exactly how it is calculated here.

The calculation looks fairly complex, but the ideas behind it are pretty simple. The basic idea is a rigorously improved version of the dumb "efficiency" stat, which simply adds up pts, rebs, etc. and subtracts missed shots and turnovers. PER does a similar thing, but multiplies the stats by coefficients based on their importance (the coefficients, I believe, come from Hollinger's regression analyses). It then adjusts for minutes played, pace, and era.

__________________
A man said to the universe:
"Sir I exist!"
"However," replied the universe,
"The fact has not created in me
A sense of obligation."

- Stephen Crane

durvasa

Does Hollinger use regression to get the coefficients? Maybe for the 0.44 on free throws,, but the rest seems to just be "common sense". For example, a defensive rebound is worth the likelihood the opponent would have rebounded the ball times the expected points they would score in a possession.

I know that Wins Produced (from the "The Wages of Wins" book) is a linear player rating formula that uses regression for the coefficients. Not sure about PER, though.

Diophantos

Yeah, I believe you're right. The example that popped into my head was the 0.44 and I didn't really think through the other ones.

__________________
A man said to the universe:
"Sir I exist!"
"However," replied the universe,
"The fact has not created in me
A sense of obligation."

- Stephen Crane

OneBadLT123

sweet, thanks man! I am starting to understand it a lot better now thanks to the link!

"You must spread some Reputation around before giving it to Diophantos again."

Dang it...Ill try gettin you some when i can

__________________
-
OneBadLT123's favorite 07-08 Rockets Videos:
41 Years of Rockets Tradition
Game # 22 in a Row Won
Saluting the Rockets
OneBadLT123's All-time Favorite Rockets Videos:
The Best Hakeem Mix
Hakeem Owning Robinson 95
Hakeem Owning Robinson Again
Post Game 95 Finals
Johnson At The Buzzer!
Tmac 13 in 33
Random Vids:
NBA on NBC Memories
Vince Carter 2000 Slam Dunk Contest

Mrobison

Maybe it's not there, maybe I'm lazy, but how does the "PER" work account for the lack of block and steal data from the 60's (and elsewhere)?

Premier

From the "Calculating PER" article at B-R.com:

Problems arise for years prior to 1980: .....1980 - debut of 3-point shot in NBA .....1978 - player turnovers first recorded in NBA .....1974 - player offensive rebounds, steals, and blocked shots first recorded in NBA The calcuation of uPER obviously depends on these statsitics, so here are my solutions for years when the data are missing: .....Zero out three-point field goals, turnovers, blocked shots, and steals. .....Set the league value of possession (VOP) equal to 1. .....Set the defensive rebound percentage (DRBP) equal to 0.7. .....Set player offensive rebounds (ORB) equal to 0.3*TRB.

__________________

Mrobison

I saw that, but I guess I missed the implication. Are you saying that, overall, that balances out (obviously, not in every individual case)? So that the whole league would have basically the same average with each system?

Premier

Yes. PER should be used to compare players each year. It is less effective when used to compare players from one era against players from another era.

__________________

Mrobison

Yep. Sadly, that's true of virtually any way to compare.