Offensive PER's of the Rockets

15 is set as the NBA average (though since better players play more, the average NBA player is below that).

For WS/48 .100 is set as average.

 

Not sure what the thread demonstrates, but I see the value in having composite pace-adjusted and minute-adjusted stats. Which isn't to say they can't be misused.

 

Plus can anyone tell why the OP did a "weighted value" on individual PERs to calculate a team PER.

I'm pretty sure PER is already weighted to mpg. So unless the OP is weighting on games played, then he is inventing his own stat based on PER

 

Exactly. Which would probably show the rockets are way above their W/L performance vs PER. Probably becuase PER yet again favors taking more shots

See

 

You are very confused. PER is a per minute stat, so the only acceptable way of taking an average of multiple players is if you take the weighted average to minutes.

 

It does tend to favor volume shooters a bit. It adjusts for pace, though, so I'm not sure the "taking more shots" criticism works on teamwide basis. I do prefer WS/48. But only rarely is there a big difference between them. (I did notice that there is a huge difference in Brandon Jennings performance with those two stats this year: he looks great with PER and only average with WS/48. Didn't try to figure out why.)

 

kevC: Can you help me out about my PPS question?

 

You think so. There's mathematical flaws in that.

The only way to do a team PER is to do one. How efficient is the teams boxscore as if it was a 48 minute player

You can can't apply weighted math to an already weighted individual stat to come to a team PER value. Show me Hollinger doing that

You will merely calculate PER to the entire teams boxscore. And if you do that then you should come close to 15 as a league average team PER

 

You guys are complicating this. A teams PER is their entire boxscore

Don't try to turn run a mpg individual stat through some further formula. Just use the entire boxscore and enter it into HOllingers stat as a single player

 

I actually agree with you on this.

 

Read the first sentence under "Introduction" section.

 

I don't know if you know what "weighted" means in this context. PER is a calculation of what the player has/is expected to produce in a given minute. Therefore, to take an average of those numbers, it has to be weighted by minutes played.

For example:
If a player A has a PER of 10 and has played 10 minutes and player B has a PER of 20 and has 20 minutes the weighted PER of those players are ((10*10 minutes)+(20*20 minutes))/(10 minutes + 20 minutes) = 16.67

If you do it for the entire team, it converges to a somewhat meaningful stat.

 

This is true, but doing it with weighted average would give you the same thing and actually probably a lot easier.

 

Of course, you actually don't need a team PER. You can just use point differential. The point of PER is to find an overall general stat for individual players. Partner, I do get your criticisms of the thread, just not of PER.

 

It seems like PPS = PPG/FGA

So the better FG% and more FT, the better PPS

 

I think the easiest thing is to just run the teams boxscore through Hollingers stat. How is that not easier?

I'd be curious to see if we run that is the league ave is indeed 15. I mean. He came up with 15 somehow

Weighting PER to a team value seems as flawed as weighting individual +/- stats to try to come up with real team pt differentials. Yeah maybe there is math to do that. But just run the team as a stat. And not try to weight the individual performance

 

Points per shot is exactly that. Points scored/number of shot attempts. This does not account for FT attempts and therefore overvalues FT's since it's just bonus points basically. TS% is (Points*100)/(2*(FGA+.44*FTA)). As you can see it accounts for free throw attempts and the .44 constant accounts for and 1's and technical FT's.

 

Actually, Meh, I think I misunderstood what you were saying. I had thought you were saying it wasn't minute adjusted--in other words that players who got more points, rebounds, etc. had higher PERs regardless of how many minutes they played. But now I'm pretty sure that you were saying that since it was minute-adjusted, one would have to weight the PERs of the individual players based on how many minutes they had played to get a team PER. Right?

Sorry about my confusion.

 

Because PER is super complicated with all these different weights (different than the minutes weighting we're talking about, it has all these approximations of what the league average is, what empirically is the negative effect of a turnover vs. a missed shot, etc etc). Hollinger or basketball-reference (who actually differ slightly because of constants and assumptions used) have already done all the work for you to boil it down to a per minute number. If one wanted to attempt to calculate a team PER the obviously easiest way is to do (PER*weighted minutes)/total minutes.

I'm not saying that this team PER concept is particularly meaningful, since there are way more insightful team metrics but if one wanted to calculate it, using weighted minutes would be the way to do it.

 

OK, I get it. I don't think it's a very useful stat then. TS%*2 is much more indicative of a shooter's real possession efficiency.