Pioneer10

I'm looking at www.82games.com vs knickerblogger both are updated to reflect the games played up to the time of this post:

82games.com: Lebron PER 26.4
Knickerblogger 24.74

Others
Wade
82 - 30.2
knick - 28.08

Dirk
82- 29.5
knick - 27.42

Seems like 82games consistenlty gives an extra 2 to the PER.

On basketballreference.com it seems to correlate much better with knickblogger stats from previous seasons

__________________

carlos710

82games stats seems inflated then. A per of 30 would have to be a really huge season. I dont think wade is having that kind of season (he has been great, but not on that kind of level imo)

Goubot

Knickerblogger has values closer to Hollinger's at ESPN, though not always exact. 82games' values are always a little high for some reason.

Brandname

Yeah, this is a bit of an oddity.

I was under the impression the PER was a trademarked statistic developed by John Hollinger. As such, it should always give the same number, and if any other methodology is used, it can't be called PER.

But I've seen at least two, possibly three different values of PER for a given player for the same span of games. I don't get it.

Premier

82games is inaccurate. They're usually slower. Knickerblogger updates daily, usually. PER is really finalized at the end of the season, though. League statistics, like pace, are easily calculated then. But, in-season KnickerBlogger is about as precise as you can get.

__________________

Theonee

None of them are, You have to actually watch the games to know how valuable a player is. PERs are just guidelines.

Brandname

I know what you mean, but that's not what you're saying.

arhie

PER is useless unless you compare it with a players Usage rate. When players have similar usage rates then you can make good valid comparisons to who had a better season. It is quite obvious that it is easier to help your team when the ball is in your hands than it is when it isn't. The PER is skewed more for Centers, and Powerforwards, as well as guards who steal the ball. PER is the most accurate for the shooting guards and small forward positions.

JPSeraph

IIRC, Hollinger himself tweaked the formula for PER slightly, leading to variations between his current ratings, and say, the ratings on basketball-reference.com - this can be seen by visiting Hollinger's site and seeing his ratings for all time players and then viewing the same on the aforementioned reference site which claims to use his formula. For instance, on Hollinger's site, Jordan's 88-89 season comes out to 32+ PER while it is just under 32 on the site...leading me to suspect they are using Hollinger's older formula.

Then again, I have never read Hollinger's book and am saying this without being aware of what was supposedly changed!

__________________
On a given day, in a given circumstance, you think you have a limit...

anh2173

Another small tidbit is the source of the data. 82games is using play-by-play data and Knickerblogger, at least in season, may be getting the data from Doug's Stats, for instance. Basketball-Reference uses the Lahman dB (I think.)

Aaron Barzilai's site has up-to-date play-by-play data for download. He also has the play-by-play back to the 2005-06 season.

BTW, Justin Kubatko (Basketball-Reference) plans to offer updated in-season stats starting some time in December. Here is a sample of the proposed new layout.

Player Page
Team Page

EHL

^ Hey that's great to hear. Looks good.