APBRmetrics

[Mountain]

I thought it might be useful to have a thread for mentioning searches folks do using this engine.

I queried 20 pt a game scorers by TS% and got these results:

http://tinyurl.com/2b55xn

Name TS% PTS
1 A Stoudemire 0.637 20.4
2 Kevin Martin 0.614 20.2
3 Dirk Nowitzki 0.605 24.6
4 Yao Ming 0.601 25
5 Pau Gasol 0.593 20.8
6 Carlos Boozer 0.588 20.9
7 Rashard Lewis 0.587 22.4
8 Michael Redd 0.586 26.7
9 Dwyane Wade 0.583 27.4
10 Elton Brand 0.581 20.5
11 Kobe Bryant 0.58 31.6
12 Tim Duncan 0.579 20
13 Chris Bosh 0.577 22.6
14 Ben Gordon 0.572 21.4
15 Paul Pierce 0.571 25
16 Gilbert Arenas 0.565 28.4
17 Ray Allen 0.564 26.4
18 Vince Carter 0.559 25.2
19 Joe Johnson 0.558 25
20 Carmelo Anthony 0.552 28.9
21 LeBron James 0.552 27.3
22 Kevin Garnett 0.546 22.4
23 Allen Iverson 0.54 26.3
24 Zach Randolph 0.537 23.6
25 Baron Davis 0.53 20.1
26 Tracy McGrady 0.515 24.6

The averages were 26.4 yrs old and almost 6'8 with perimeter and bigs about proportionally represented.

TS% 57.2 with overall FG 48.1% 3pt 31.6%. Getting to the free throw line helps the TS% of these guys alot.



Checking 8+ a game rebounders I found about 40% were 6-9 or shorter.

[Statman]

[Mountain]

PER 15 is set at league average performance level but only 132 of 468 players achieved it last season, or about 28%. 52% were below 13. 80 (or about 17%, influenced by short-termers and very light deep bench players) were below 8.

Should 13 (or more precisely 12.8 ) be thought of as "a new 15"? If / when you mean "average player" instead of average performance level?

[DLew]

Statman,

I don't know what you mean by reward. Scoring from the line is very high percentage, hence it drives up TS%, but it is not rewarded at all. Free throw scoring is evaluated the exact same way as 2 pt scoring and 3pt scoring: points divided by possessions used. There is no reward there.

My argument was, that given two players, one that uses 10 possessions and scores 14 points by shooting all 2s (by going 7 for 10 for example), and one that uses 10 possessions and scores 14 points by shooting all free throws (by going 14 for 23 on 7 two shot fouls and 3 three shot fouls for example), you would prefer to have the free throw guy even though their TS percentages would generally be equal. This is because of the extra effects that drawing fouls have.

So, as you can see TS% gives no reward for the second order effects of getting to the line, it simply properly values the first order effects, which is its purpose.

[John Hollinger]

Mountain,

you bring up a good point -- the league average PER is based on all minutes by all players. But of course, the best players play the most minutes, so invariably the result is skewed by 3,000 minutes of LeBron James as compared to 1,000 of Jannero Pargo. That's why players with PErs is low teens can still be really helpful.

[Statman]

[Mountain]

Right John, a player with a low teens PER can still be helpful especially on a good team with a strong top 2-3 on scoring and especially if they have a skill strength that helps balances out the 4 factors for the team on offense and defense. Lots of vital players on the best teams with low teens PER.

[gabefarkas]

[Mountain]

Yes it is, as John states "the league average PER is based on all minutes by all players".

I got used to thinking of PER15 as average in a loose way and it is average level of performance; but I thought it was worth noting it isn't average player quality for whole league.

It is closer to average of rotation players. Only 260 players played 1000+ minutes. 112 or 43% were PER15+. 175 or 67% were PER 13+. Less than 10% were lower than PER10, including 8-10 starters.

(So 20 players lower than 1000 minutes were also PER 15+. About a third were regulars affected by injury, half were rookies and there were just a few vets who just have limited roles due to roster depth or coaching preference.)

Some PER 11 - 14 players deserve a little more respect. It is still weak for those who start, though sometimes one or two starters in the shadow of a dominant offensive player will slip down there due in large part to limited supply of shots.

[deepak_e]

Maybe it would be a good idea to come up with a PER variant ("PER2") which rates players relative to the median rating, rather than the average rating (weighted by minutes played).

[Mountain]

A calculation of PER/median PER for that year is easy enough to assemble adhoc (using a database to locate the median). A PER 14 last season / roughly 12.8 median PER = 1.094 or multiply by 100 and say 109.4 if that form at is more appealing. If done globally maybe it could be called PERQ (for relative quality compared to median score) or some such.

[Mountain]

http://tinyurl.com/3484gb

Name FGA per 48 minutes
Carmelo Anthony 28.1
Tracy McGrady 28
Kobe Bryant 26.9
Zach Randolph 25.5
Gilbert Arenas 25.3
Ray Allen 24.9
Vince Carter 24.5
LeBron James 24.4
Yao Ming 24.2
Michael Redd 24.1
Dwyane Wade 23.9
Ben Gordon 23.8
Paul Pierce 23.4
Joe Johnson 23.2
Allen Iverson 22.8
Dirk Nowitzki 22.8
Willie Green 22.6
Baron Davis 22.3
Jermaine O'Neal 22.3


Only 8 players take 20 FGAs per game

http://tinyurl.com/2kb3dt

Name FGA
Kobe Bryant 22.8
Carmelo Anthony 22.4
Ray Allen 21
Gilbert Arenas 20.9
LeBron James 20.8
Tracy McGrady 20.8
Allen Iverson 20.2
Joe Johnson 20

[Mountain]

http://tinyurl.com/323xfn

Sam Cassell was the oldest player with a PER over 17, barely, at age 37. Using a 1000 minutes played qualifying standard there 15 other 30 somethings. 28 between 25-29, 24 under 25. A total of 69. None under 21. 6 at 21. You could compute %s of populations as well.

http://tinyurl.com/2tm9rl

5 7 footers within the 20 6'10+, 20 between 6'7 and 6 '9, 24 6'3 to 6'6, only 5 6'2 and under. If you go by roster heights. Yao Ming the tallest, Chucky Atkins the shortest. (An argument against drafting guards listed 6'2 and under or counting on them big on offensive side of ball or paricularly as scorers?)

http://tinyurl.com/3992sr
Garnett and Odom only players over 6'8 with 4 assists a game though 40% of the 56 such players were over 6-5 and only a few these played some PG. That only leaves 1 traditional point sized guard with this level of production per team.

http://tinyurl.com/3amjf9
Among 71 players with 2.0+ blocks per 48 only 5 were under 6'9- Balkman, Milsap, Brand, Singleton, Maxiell.

http://tinyurl.com/2oevrc
Among 85 players with 3+ 3 pt attempts a game 26 were 6-8+, 34 were 6'4 to 6'7 and 25 under 6'4. The tallest was Okur or Bargnani depending on how you report the later's height. Enough for one per team per height on average. The team distributions by height could be another angle for evaluating 3 pt attack strategy and capability.

Trivia but perhaps some will be of interest to a few.

[jkubatko]

[Mountain]

I considered posting the link (Almost did but happened not to; no slight intended.) but simplified the presentation for easy of reading and try to ease possible start of discussion around a summary finding (to save folks time) because I believe fewer people follow links than read what is right on the forum. (Tried.) But I don't mind giving the links.

Here are a few more:

Only 29 players 30+ years / playing 2000 minutes last season by VAR
http://tinyurl.com/yrte9u

Only 85 players averaging 5+ FGM a game by pts avg.
http://tinyurl.com/yvhq5u

23 players averaging 5+ FT made a game
http://tinyurl.com/2a8oke

46 player over 5FTA a game by age
http://tinyurl.com/2b8zsb
17 25 and under, only 8 over 30.

It is a helpful tool that I was highlighting. Now with links.