Jamesian blog

[mtamada]

Lebron and Bill, that is. An apbrmetrical send-up of things that LBJ might say, if he knew advanced hoopstats. Don't know who's behind it.
http://billjameslebronjames.tumblr.com/

[huevonkiller]

They should have this on the site:

http://bkref.com/tiny/voZmn

LeBron's peak seasons are the best of all-time. He's unmatched in his playoff prime.


http://bkref.com/tiny/ry8le

Wow, he might be underrated.

[agentkirb]

I can respect the theory behind statistics like wins-shared. But when you look at a list that has Marcus Camby, Chris Paul and Amare Stoudimire both in the top 15 ALL TIME... I really have to wonder what they are measuring.

[huevonkiller]

Amare had almost 30 PER so your criticism of that seems unfounded. The key is to look at both PER and Ws/48. Jordan in 96 should probably be lower.

Hmm the second link addresses statistical noise. WS/48 is probably a better stat than PER most of the time, because of the defensive component and adjustment for usage.

I can respect the theory behind statistics like wins-shared. But when you look at a list that has Marcus Camby, Chris Paul and Amare Stoudimire both in the top 15 ALL TIME... I really have to wonder what they are measuring.

A perplexing comment, since Chris Paul at his peak is better than many of Jordan's seasons.

[Mike G]

Chris Paul's peak PER is 30.0, and his peak WS/48 is .292, both in 2008-09.
Among guards since 1987, that's the 5th best WS rate among guards (2000+ minutes); Jordan has 7 of the top 11, and Paul has the other 4.

Jordan also takes the top 4 PER in the era, with Wade bumping Paul to 6th in that stat, also for 2009.
You also get quite a different list of players if you require 600 playoff minutes, or 150.

Paul isn't the most questionable inclusion in that list; Camby probably is. His 1999 playoffs -- ranked #7 in WS/48 -- behind 3 Jordans, 2 LeBrons, and 1 Paul -- rank higher than Duncan '03, Magic '86 (#12), Bird '86 (#15), Moses '83, Shaq '01 ...

It does stretch credibility (or is it credulity?)
Camby's PER for that postseason ranks just 58th. That's still pretty elite, but it supports the notion that a blend of PER and WS/48 would generally be a better gauge of per-minute effectiveness than either used alone.

[Mike G]

Actually, this was looked at after last year's playoffs:
viewtopic.php?f=2&t=288

Ranked by the quantity (WS/48)*WS*PER^.5
(that is, the square root of PER)
This gives more credit to more minutes, via total Win Shares.

Done year by year, some credit should perhaps also be given to annual leadership/ranking. Some years, there are just more playoff games.

[agentkirb]

A perplexing comment, since Chris Paul at his peak is better than many of Jordan's seasons.

I haven't completely looked into the science behind win shares. I imagine it's some measure of how many "wins" a player was worth for their team. And I can respect such a measure because they don't want to just look at points, rebounds, assists and FG%. But it's kind of hard for me to separate the stats from the tape when you compare a guy like MJ who is obviously as accomplished as he is... being in same GOAT group as a guy like Chris Paul who hasn't made it further than the 2nd round. And I know the Hornets weren't that great of a team other than Paul, but they were as good of a supporting cast as the Magic and the Cavs when they both made it to the finals. And they both beat legit teams to get there.

I think it's a good idea to try to achieve a statistic that really gets to the bottom of which players are better. But certainly no statistic is going to be perfect and at the end of the day you have to look with your eyes and go "ok, that doesn't make any sense".

And obviously Camby and Amare are more suspicious than Paul, but you aren't going to hear anyone put Paul and MJ in the same GOAT conversation.

[bchaikin]

but you aren't going to hear anyone put Paul and MJ in the same GOAT conversation...

of course you are...

many consider jordan to be the greatest SG to ever play the game, possibly the greatest player period. others may consider magic, bird, chamberlain, or russell the greatest player ever, but they played different positions than jordan did...

paul certainly has a chance to be one of the greatest PGs of all time, maybe the greatest ever once his career is done. but his career isn't done yet - he's only 27 years old. just because he hasn't won any titles doesn't preclude him from the discussion. karl malone and charles barkley never won titles, but each is considered an all-time great...

paul's been in the league for just 7 seasons. the last 5 seasons his stats have been outrageous for a PG - 20 pts/g, a 57.4% ScFG% (2s, 3s, and FTs), 10+ ast/g, just 2.5 TO/g, and more steals than turnovers, all-NBA 1st team twice, all-NBA 2nd team once, all-D 1st team twice, all-D 2nd team twice. not many PGs can claim all this by the age of 27...

after having played just 7 seasons both malone and barkley hadn't won any titles, but both were still considered future hall-of-famers. by the age of 27 jordan had yet to win a title (he was 28 when he won his 1st), but was already considered a great player and a future for the hall-of-fame...

[Mike G]

In several ways LeBron is ahead of Jordan at the same age:
http://bkref.com/tiny/1KMay

In regular seasons, MJ leads LeBron in PER, 30.2 to 27.2; in WS/48 by .275 to .233
And in playoffs, the difference is almost identical. Both have retained their normal rates in the postseason. That is to say, both have been great in playoffs.

LeBron has been in the league 2 years longer at age 27. Remove his first 2 seasons, and their career PER and WS/48 are a lot closer.
LeBron has played 40% more minutes in RS, and he has 70% more playoff minutes. He's been to the Finals 3 times vs once by Jordan.

Jordan had 2 MVP's by age 27, and LeBron has 3.
At 26, Paul is right behind Jordan in RS Win Shares (in 8% more minutes), and just ahead of LeBron in WS/48, all with zero titles.
http://bkref.com/tiny/hW9hA

[Mike G]

Using the formula xyz = (WS/48)*WS*PER^.5, best postseasons since 1978:

Code: Select all

xyz     Player          yr    Tm    G   Min    WS    WS/48   PER
11.8  LeBron James     2009   Cle   14   41   4.82   .399   37.4
9.1   LeBron James     2012   Mia   23   43   5.82   .284   30.3
9.0   Michael Jordan   1991   Chi   17   41   4.78   .333   32.0
8.8   Tim Duncan       2003   SAS   24   43   5.93   .279   28.4
7.4   Michael Jordan   1996   Chi   18   41   4.67   .306   26.7

7.3   Dirk Nowitzki    2006   Dal   23   43   5.39   .263   26.8
6.8   Michael Jordan   1998   Chi   21   42   4.81   .265   28.1
6.5   Michael Jordan   1993   Chi   19   41   4.40   .270   30.1
6.4   Michael Jordan   1990   Chi   16   42   3.99   .284   31.6
6.0   Dwyane Wade      2006   Mia   23   42   4.80   .240   26.9

xyz     Player          yr    Tm    G   Min    WS    WS/48   PER
6.0   Michael Jordan   1989   Chi   17   42   4.04   .270   29.9
5.8   Shaquille O'Neal 2000   LAL   23   43   4.67   .224   30.5
5.7   Kobe Bryant      2009   LAL   23   41   4.66   .238   26.8
5.7   Larry Bird       1984   Bos   23   42   4.72   .236   26.3
5.5   Dwight Howard    2009   Orl   23   39   4.53   .241   25.5

5.4   Larry Bird       1986   Bos   18   43   4.22   .263   23.9
5.4   Manu Ginobili    2005   SAS   23   34   4.18   .260   24.8
5.1   Shaquille O'Neal 2001   LAL   16   42   3.66   .260   28.7
5.0   Magic Johnson    1987   LAL   18   37   3.68   .265   26.2
4.9   Charles Barkley  1993   Phx   24   43   4.60   .215   24.9

Because WS are dependent on minutes, we miss playoffs like Moses' (fo-fo-fo) near sweep in '83; and in fact the whole field is biased toward later years, now with 4 full rounds for finalists.

We can minimize the minutes factor a bit by taking the square root of WS.
xyz2 = (WS/48)*(WS^.5)*(PER^.5)

Code: Select all

xyz2     Player          yr     Tm   G   Min    WS    WS/48   PER
5.36   LeBron James     2009   Cle   14   41   4.82   .399   37.4
4.12   Michael Jordan   1991   Chi   17   41   4.78   .333   32.0
3.77   LeBron James     2012   Mia   23   43   5.82   .284   30.3
3.62   Tim Duncan       2003   SAS   24   43   5.93   .279   28.4
3.42   Michael Jordan   1996   Chi   18   41   4.67   .306   26.7

3.19   Michael Jordan   1990   Chi   16   42   3.99   .284   31.6
3.16   Dirk Nowitzki    2006   Dal   23   43   5.39   .263   26.8
3.11   Michael Jordan   1993   Chi   19   41   4.40   .270   30.1
3.08   Michael Jordan   1998   Chi   21   42   4.81   .265   28.1
2.97   Michael Jordan   1989   Chi   17   42   4.04   .270   29.9

xyz2     Player          yr     Tm   G   Min    WS    WS/48   PER
2.74   Hakeem Olajuwon  1988   Hou    4   41   1.30   .385   39.0
2.74   Chris Paul       2008   NOH   12   41   2.93   .289   30.7
2.73   Dwyane Wade      2006   Mia   23   42   4.80   .240   26.9
2.67   Shaquille O'Neal 2000   LAL   23   43   4.67   .224   30.5
2.67   Shaquille O'Neal 2001   LAL   16   42   3.66   .260   28.7

2.66   Kobe Bryant      2009   LAL   23   41   4.66   .238   26.8
2.65   Manu Ginobili    2005   SAS   23   34   4.18   .260   24.8
2.64   Larry Bird       1986   Bos   18   43   4.22   .263   23.9
2.63   Larry Bird       1984   Bos   23   42   4.72   .236   26.3
2.60   Magic Johnson    1987   LAL   18   37   3.68   .265   26.2

Now Jordan is at #2 with his much better PER and WS/48 rates than LeBron's this year. Big difference is that the '91 Bulls finished off their opponents more swiftly.
We also get Olajuwon's 4 games in 1988. As strong a postseason as any, but very brief.

[Mike G]

It's a desperate measure; but if we just use a hard cutoff for minutes -- 500, say -- we can give equal weight to all eras.
xy = (WS/48)*PER

Code: Select all

.xy.      Player          yr     Tm   G   Min    WS    WS/48   PER
14.92  LeBron James      2009   Cle   14   41   4.82   .399   37.4
10.66  Michael Jordan    1991   Chi   17   41   4.78   .333   32.0
8.97   Michael Jordan    1990   Chi   16   42   3.99   .284   31.6
8.61   LeBron James      2012   Mia   23   43   5.82   .284   30.3
8.17   Michael Jordan    1996   Chi   18   41   4.67   .306   26.7

8.13   Michael Jordan    1993   Chi   19   41   4.40   .270   30.1
8.07   Michael Jordan    1989   Chi   17   42   4.04   .270   29.9
7.92   Tim Duncan        2003   SAS   24   43   5.93   .279   28.4
7.46   Shaquille O'Neal  2001   LAL   16   42   3.66   .260   28.7
7.45   Michael Jordan    1998   Chi   21   42   4.81   .265   28.1

.xy.      Player          yr     Tm   G   Min    WS    WS/48   PER
7.06 Kareem Abdul-Jabbar 1980   LAL   15   41   3.26   .253   27.9
7.05   Dirk Nowitzki     2006   Dal   23   43   5.39   .263   26.8
6.94   Marcus Camby      1999   NYK   20   25   2.97   .280   24.8
6.94   Magic Johnson     1987   LAL   18   37   3.68   .265   26.2
6.84   Magic Johnson     1986   LAL   14   39   3.01   .267   25.6

6.83   Shaquille O'Neal  2000   LAL   23   43   4.67   .224   30.5
6.68   Moses Malone      1983   Phl   13   40   2.84   .260   25.7
6.68   Shaquille O'Neal  2002   LAL   19   41   3.82   .236   28.3
6.50   Kobe Bryant       2001   LAL   16   43   3.76   .260   25.0
6.46   Dwyane Wade       2006   Mia   23   42   4.80   .240   26.9

We got Moses, and all 3 of Shaq's Lakers titles; but also that damn Camby.
If I'd cut minutes at 486, Chris Paul '08 would be #4 in this list!

[huevonkiller]

Lebron had deep playoff runs when he was 23 and younger, so that probably underrates his career playoff stats. His ws/48 in his first three seasons are the same as Jordan's 21-23. LJ's peak is incredible, he's clearly got the most potential of any player ever.

Chris Paul's peak PER is 30.0, and his peak WS/48 is .292, both in 2008-09.

Jordan has a 31 PER season where he spams usage rate (a little bit) and therefore has a .285 WS/48. Chris Paul's peak season is better than most of Jordan's seasons.

Per must be adjusted for usage and defense. Ws/48 seems to adjust for this well, but not always. Chris Paul is underrated.

Among guards since 1987, that's the 5th best WS rate among guards (2000+ minutes); Jordan has 7 of the top 11, and Paul has the other 4.

Jordan also takes the top 4 PER in the era, with Wade bumping Paul to 6th in that stat, also for 2009.
You also get quite a different list of players if you require 600 playoff minutes, or 150.

Jordan only has 3 seasons better than 09 Paul, and 1 other year where he is on-par with 09 Paul. I don't count 1996 because his PER is lowish compared to prime LeBron/Jordan.

09 Paul's defensive rating is inferior to LeBron and Jordan, but he probably played better than 09 Wade.

Paul isn't the most questionable inclusion in that list; Camby probably is. His 1999 playoffs -- ranked #7 in WS/48 -- behind 3 Jordans, 2 LeBrons, and 1 Paul -- rank higher than Duncan '03, Magic '86 (#12), Bird '86 (#15), Moses '83, Shaq '01 ...

It does stretch credibility (or is it credulity?)

Camby's PER for that postseason ranks just 58th. That's still pretty elite, but it supports the notion that a blend of PER and WS/48 would generally be a better gauge of per-minute effectiveness than either used alone.

Yeah I use both.

[Mike G]

After a phenomenal 2008-09 season for Chris Paul, his Hornets went down in flames in the playoffs, losing to Denver by some 25 ppg.
Check out the Playoffs Advanced section: http://www.basketball-reference.com/teams/NOH/2009.html

Paul had their only above-avg PER (16.1); then Posey at 13.3; and West, 12.2
Their WS/48 were respectively -.035, -.004, and -.071

I'm curious what anyone thinks of these PER and WS rates. Is one more reliable than the other for such sample sizes? What accounts for these enormous disparities?

Just a couple dozen players in this century have had a season WS/48 rate as low as West's (-.071) in this series, in as many (178) minutes.

[huevonkiller]

You still have the same problem. Chris Paul's 08-season when you count the playoffs, is better than all but 3 or 4 of Jordan's seasons.

After a phenomenal 2008-09 season for Chris Paul, his Hornets went down in flames in the playoffs, losing to Denver by some 25 ppg.
Check out the Playoffs Advanced section: http://www.basketball-reference.com/teams/NOH/2009.html

Paul had their only above-avg PER (16.1); then Posey at 13.3; and West, 12.2
Their WS/48 were respectively -.035, -.004, and -.071

I'm curious what anyone thinks of these PER and WS rates. Is one more reliable than the other for such sample sizes? What accounts for these enormous disparities?

Just a couple dozen players in this century have had a season WS/48 rate as low as West's (-.071) in this series, in as many (178) minutes.

And yet in spite of all that, the year before he had a 30.7 PER and .289 WS/48. And he had a .284 WS/48 for the entire 07-08 season. His Prime is still better than most of Jordan's career, even if you take into account the playoffs.

The biggest problem Chris Paul has had over his career is his knee/groin problems. He's not the most durable guy.

[Mike G]

Understood that Paul is an elite player.
Even in a bad series, his PER was decent, at 16
For the same series, his WS/48 was roughly that of a replacement-level player, for some 200 minutes.

My question is: Does either metric have a tendency to widely miss a player's actual value, under extreme conditions -- as in, great team success or failure?
If so, are there reasonable "corrections" that could be made?