APBRmetrics

1.Jordan
2.Russell
3.Wilt
4.Kareem
5.Magic
6.Bird
7.Lebron
8.O'Neal
9.Duncan
10.the predator
11.Oscar
12.West
13.Hakeeeem!!
14.Doc
15.Pettit
16.Mikan
17.Moses
18.Cousy
19.Mailman
20.Baylor
21.Havlicek
22.Dirk
23.Garnett
24.Barkley
25.Wade

If it can be quantified with numbers, then it can be used as a stat. That includes MVPs, All-NBA selections and, yes, championships.

Based on box score stats alone:

1.Wilt
2.Jordan
3.Kareem
4.Lebron
5.Oscar
6.Magic
7.Bird
8.Shaq
9.Mailman
10.Barkley
11.D-Rob
12.Pettit
13.West
14.Hakeem
15.Duncan
16.KG
17.Stockton
18.Dirk
19.Kobe
20.Russell
21.Doc
22.Gervin
23.CP3
24.Moses
25.Wade

AYC: You realize that to vote, you have to check candidates' names in the Poll ?
Many of your names have been selected already, so are not still available.
I haven't heard of "the predator".

Voting will conclude Tues. morning (tomorrow) as usual.
Someone dropped in with 3 votes recently.

Just a brief, tangential follow-up to Mike's challenge:

Mike G wrote:Rather than construct arguments designed to put your guy at the top, and convince yourself that it's infallible; maybe try the exercise of picking Any player from the list and making such an argument. It's entirely do-able.

When this "Best Of..." string came up, I was very skeptical that any meaningful statements could be made about identifying the best players of all time. However, having pored over and contemplated all available +/- type data, in particular Jeremias' real and "fake" xRAPM (with valuable context provided by Eli Witus' 2006 positional averages for APM) as well as the "gross" +/- of team ORtga and DRtg, I think there is a very robust result as to who are the four best players to have started their careers since 1980, with some question as to the ordering within the group.

In likely descending order of superiority: David Robinson, Shaquille O'Neal, LeBron James, and Tim Duncan. After that there is a step down with perhaps Kevin Garnett heading the group, but things get a bit more speculative in identifying its members.

The only controversial point regarding the ordering of the first group is linking "fake" to real xRAPM. Average fake RAPM for elite players (rank by rank) is significantly below real, over two points for the top three positions (2.3, 2.5, and 2.5, in order), descending to 1.7 for the fifth best player, and 1.2 for the tenth. And there is a clear discontinuity between 2000 and 2001. Making such "naive" adjustments allows us to recreate David Robinson's career ratings, and what a career it is shown to be!

A secondary point to note is that LeBron James' career is obviously not yet nearly over, and on a year by year comparison, he's not that far behind Shaq or that far ahead of Duncan, so he may or may not change rank. He is far enough behind David Robinson, however, (again, given the aforementioned disclaimer) that LBJ can only aspire to being number two.

As for "where's Jordan, Bird, and Magic", well, as previously noted, if you don't play center or a center-like defensive position (or your name isn't LeBron) you face a huge disadvantage in being eligible for being considered the best of all time. Being tall, a good defender, and having responsibility for protecting the basket simply gives you a huge advantage where it counts (effecting positive change on the scoreboard).

Most surprising perhaps is Jordan, where his rank could well be below Bird and Magic. But why should it be surprising? There is, again, an adding-up constraint that must be observed. And in the same spirit as the exercises down with Bill Russell, if Jordan was so great, who were the counterpart bad players, for example, on the 1996 Bulls? In this light, the "fake" xRAPM makes perfect sense.

Similarly though the direct evidence is thin, contextual evidence based on team performance and lack of contemporary analogs, very strongly suggests that Bird and Magic couldn't have been "that" great. "That", again, being defined to include several +10 seasons.

Finally, there is the issue of whence Wilt and Kareem? Here the evidence is entirely contextual, but I am far more down on these players than when I began all this. On the face of it, the prior should be that given their offensive and defensive reputations and the advantage that offensively and defensively skilled centers have given the realities of basketball that they should surely be in the Top 5 (or 6). But, frankly, having looked over the entirety of their "on/offs" (i.e. when the joined and left various teams) in addition to discrete changes in their team's performance levels and what explained these, there is of course evidence that they were very good (in particular, their impact on their teams in their rookie seasons) but after this there is not much at all that suggests singular greatness.

These points can be discussed in greater detail if anyone wishes, but for now, these two players must remain asterisks in my mind.

Mike G wrote:AYC: You realize that to vote, you have to check candidates' names in the Poll ?
Many of your names have been selected already, so are not still available.
I haven't heard of "the predator".

Voting will conclude Tues. morning (tomorrow) as usual.
Someone dropped in with 3 votes recently.

Nor had I. I figured by a vague association with a more known (real?) nickname (Black Mamba) and checked by process of elimination that he means Kobe.

I thought I had voted, but I didn’t push submit. Oops. But I wanted show the order in which I would place my 25 anyway. I had a hard time putting a certain sexual predator from the Lakers in my top 10; intellectual honesty requires me to put him there, but I don’t have to write his name there.

Some thoughts on Russell, the number two guy on my list:
1) Having 3 all-stars for teammates isn't all that special when there are only 4-5 teams per conference.
2) How can we use win shares to evaluate other players and not mention Russell's Ruthian dominance on the defensive side of the ball according to that very metric?
3)Russell’s career total of 133.6DWS is 41% higher than second place Kareem, despite the fact that Kareem played in nearly 600 more games.
4)Russell’s career avg of 11.4 DWS per 82g is 61% higher than second place Wilt, and virtually equals Michael Jordan’s career avg of Offensive WS per 82g.
5) We can’t be sure DWS correctly apportions defensive credit to the individual players on a team, but given that the Celts as a team improved by 18 DWS in his 1st season and declined by 18 DWS the year after he retired, we can infer that DWS understates his defensive value, especially since weak defenders like Cousy and Heinsohn put up big DWS numbers, simply as a result of playing alongside Russell.
6) If nothing else, we know the Cs had the best defense in the league for 12 out of Russell’s 13 seasons, finishing a close second to Wilt’s Sixers in 1968

AYC wrote:Some thoughts on Russell, the number two guy on my list:
1) Having 3 all-stars for teammates isn't all that special when there are only 4-5 teams per conference.
2) How can we use win shares to evaluate other players and not mention Russell's Ruthian dominance on the defensive side of the ball according to that very metric?
3)Russell’s career total of 133.6DWS is 41% higher than second place Kareem, despite the fact that Kareem played in nearly 600 more games.
4)Russell’s career avg of 11.4 DWS per 82g is 61% higher than second place Wilt, and virtually equals Michael Jordan’s career avg of Offensive WS per 82g.
5) We can’t be sure DWS correctly apportions defensive credit to the individual players on a team, but given that the Celts as a team improved by 18 DWS in his 1st season and declined by 18 DWS the year after he retired, we can infer that DWS understates his defensive value, especially since weak defenders like Cousy and Heinsohn put up big DWS numbers, simply as a result of playing alongside Russell.
6) If nothing else, we know the Cs had the best defense in the league for 12 out of Russell’s 13 seasons, finishing a close second to Wilt’s Sixers in 1968

Re Point 1: There was a cap on all-stars per team. Otherwise Boston would likely have had many more ASG appearances. They have the All-NBA and HoF credentials too.

2-4) I suspect more staunch Russell advocates won't want to bring win shares into it. Because his advantage in that area is substantially smaller than the great advantage many of the great centers have over him on the other end. For example http://www.basketball-reference.com/pla ... 01&y6=2003

5) What's your source for Heinsohn and Cousy being bad defenders? I'm genuinely interested. Because just because a player wasn't amongst the notably strong defenders on those teams, I've seen Russell advocates throw his teammates under the bus. Not that they aren't plausibly overvalued by DWS (DWS isn't a great metric to my mind, and I haven't seen nearly enough of the Celtics to judge their ability based on that), but I would want real, preferably contemporary, sources before deciding a player (from that era) was weak on D. To the broader point we can certainly see Russell had a huge impact on D. But I'm wary of those who say we can exactly isolate what is Russell if (as at the end) there's changes in personel, player deterioration, change of coach, change of style, loss of continuity etc.

6) True. And Russell is probably the best defender of all time.

In likely descending order of superiority: David Robinson, Shaquille O'Neal, LeBron James, and Tim Duncan. After that there is a step down with perhaps Kevin Garnett heading the group, ...

No apparent consideration of playoff performance in comparing careers?
Jordan, Olajuwon, to a lesser degree Bird and Magic, were rather dominant in their postseasons. Oftentimes they clearly outplayed a 'better' opponent.
Robinson's the odd man out, in this consideration; we all know why.

Russell is probably the best defender of all time.

Is "probably" rather a strong word?
He was revolutionary in his day, but how can that be compared to players since then? And if he really didn't do much scoring, how does his D compare to someone who was a threat at both ends?

Is Rodman the best rebounder of all time? Or did he just do nothing else?
Bill Sharman led all guards in FG% several times, while leading all players in FT%. Is he one of the greatest shooters of all time?
Well, he took set shots. And he has one of the top 80 FT% seasons of all time.

Olajuwon was his team's go-to offensive weapon while also playing obliterating defense. How much better could Russell have been? Transpose those guys in a time machine, and try to declare one team got the better deal.

They actually had a similar story line. Russell got out of the deep South, where basketball was hardly played, to California. Both would become premier postseason players after regular seasons when other centers were called 'better'.

The poll has been refreshed, with 8 more players replacing 8 we've elected with 4 or more votes. We now have a Top 23 that looks like this:We've had 3 votes in this '2nd iteration', electing players 1 thru 6, 7 thru 15, and 16 thru 23.
Ties are 'broken' by 1st-iteration votes. Pettit was not elected in the first round of the 1st iteration (to the Top 25); Pippen, Havlicek, and Wade were top 25 'then', not top 23 now.

These rankings aren't written in stone by the tiebreakers I've employed. We could have run-offs if there's interest.

FWIW the faux EWA-win shares pythagorean ranking I've refered to in earlier iterations of this discussion, has all of it's top 21 in our top 23. It doesn't have Erving (though if ABA numbers were included in both sets of numbers he would be, as would Artis Gilmore) or Baylor in its top 23.

Erving's absence is explained above. Baylor is slightly disadvantaged by the faux-EWA skewing anti-older era players (he ranks 26 here), but mainly because he's 65th on Win Shares.

Anyway here's it's 22-25
Patrick Ewing
Jason Kidd
Paul Pierce
Robert Parish
Clyde Drexler
Scottie Pippen
Gary Payton
Dolph Schayes
Bob Lanier
Dominique Wilkins
Adrian Dantley
Reggie Miller
Shawn Marion
Artis Gilmore

Erving and Baylor are in the next 5 (Julius Erving, Walt Frazier, George Mikan, Elgin Baylor, Ray Allen)

Mike G wrote:

In likely descending order of superiority: David Robinson, Shaquille O'Neal, LeBron James, and Tim Duncan. After that there is a step down with perhaps Kevin Garnett heading the group, ...

No apparent consideration of playoff performance in comparing careers?
Jordan, Olajuwon, to a lesser degree Bird and Magic, were rather dominant in their postseasons. Oftentimes they clearly outplayed a 'better' opponent.
Robinson's the odd man out, in this consideration; we all know why.

First, an addition of an omission. By the same criteria that 1 through 4 were selected, Hakeem gets tucked in neatly on one side or the other of KG. Comparing available years, Hakeem is slightly ahead, but KGs two monster years by xRAPM occurred just before the comparable record.

Regarding playoff performance, I am rather reluctant to include this as an independent category for a couple of reasons. First and foremost because the +/- data I am using isn't available (or do they include playoffs?). And second, because it isn't clear to me that it is especially informative (small sample size).

Also, one can easily be led astray when thinking about playoffs, as there is a tendency to assign team credit to an individual and, of course, disproportionately weight the (hopefully) better understood offensive performances at the expense of defensive.

With this thought in mind, I went to basketball-reference to see what I might learn, taking a team perspective, about the Bird Celtics, Johnson Lakers, Jordan Bulls, Olajuwon Rockets, and Robinson Spurs (pre-Duncan), with a simple question in mind: did these teams have a habit of over or underperforming in the playoffs. Of course, there are several criteria one could choose. I chose three simple ones: (1) In non-championship years, was the loss to a superior or inferior opponent (as determined by Predicted Wins in B-R)? (2) In championship years, was a superior opponent faced? and (3) Regardless of season outcome, were any superior opponents beaten before being eliminated?

And the answers? Well, the answers are kind of as expected, but there are some surprises too, I think. Going best to worst, as expected, Olajuwon. Only twice out of 11 seasons were his teams eliminated by an inferior opponent. And with both championships, superior opponents were faced. And in all 13 playoff seasons, six times at least one superior opponent was defeated.

Arguably next comes Jordan, but I say arguably because there is no strong basis for thinking of the Bulls' team success in terms of Jordan, as opposed to Pippen, and also there weren't many years when they weren't also the best team in all of the NBA, where playoff success was implied. The count? The Jordan Bulls only lost (seven times) to superior opponents. But they never saw a superior opponent in any of their six championships. And then two times out of six when eliminated, they beat at least one superior opponent (and then when Pippen was flying solo, that year they too beat a superior opponent).

Then who comes third, by a nose, it is the Johnson Lakers (1980 to 1991), edging out (you didn't guess it) the Robinson Spurs. In the seven, non-championship years, three times Magic's team lost to an inferior opponent - the same record as the Admiral's. With championships (obviously the Spurs didn't win any) and the Lakers actually did win one out of five against a superior opponent. Then in the non-championship years, two out of seven times did the Lakers beat a superior opponent along the way, and this is what breaks the tie with the Robinson Spurs; they only beat a superior opponent one time out of seven.

And the loser, by a long shot, is.... the Bird Celtics. In his playoff, non-championship years, six out of eight times the Celtics lost to inferior opponents. One of the three championships saw them defeat a superior opponent along the way. And never in their losing playoff runs did they ever win a series against a superior opponent.

All this is kind of interesting, but important?

As for the answer regarding Olajuwon's place in the pantheon, I don't think it has much influence, at least in regards to his rank relative to Robinson. Again, conditional on my "naive" transformation of fake to real RAPM (which actually relatively benefits Hakeem) he is simply too far behind. We have directly comparable data for their ages 28 to 37, and in those years, David Robinson is +2.2 on Hakeem. And this includes a disastrous dip in Robinson's rating for his six game 1997 season. You eliminate this weirdly divergent figure from the average and the average margin over nine years is +2.9. That is a really big difference.

Mike G wrote:

Russell is probably the best defender of all time.

Is "probably" rather a strong word?
He was revolutionary in his day, but how can that be compared to players since then? And if he really didn't do much scoring, how does his D compare to someone who was a threat at both ends?

Mike G wrote:Olajuwon was his team's go-to offensive weapon while also playing obliterating defense. How much better could Russell have been? Transpose those guys in a time machine, and try to declare one team got the better deal.

I would like to encourage everyone to avail themselves of the tremendous gift that Jeremias provides at http://stats-for-the-nba.appspot.com/. Poring over the data, there is a firm context provided for testing one's preconceptions about both individual players as well as the game itself. And when a significant divergence appears between a given xRAPM player rating and a prior belief ask oneself the question: why do the gods of randomness and regression conspire against me?

I note this in regards to the juxtaposition of Russell and Olajuwon. I think all will agree that Russell, by reputation, was a great, great defensive specialist, and I like to think that now we all know what that looks like (Ben Wallace, cough, cough) and we also know, almost by definition, that such players aren't likely to make a positive contribution on offense. But what about a well-balanced player like Olajuwon?

Well-balanced? Are we really sure that he was well-balanced? Fake xRAPM paints a very consistent picture (data what dovetail with his last two seasons when there is real xRAPM): Hakeem Olajuwon was quite simply not a great offensive player. His highest rating (not prorated according to my scheme) was +1.4 in 1996, and his average fake xRAPM rating (1991 to 2000, ages 28 to 37) was just 0.1. Given the suggested age curve for the non-existent first few years of his career, it is highly implausible that he had a positive career average for Offensive xRAPM. But all our eyes tell us that he was great, so he must have had a positive impact on his team for all his offense. He simply must.

So the irony is that Olajuwon appears to have been a insufficiently-specializing defensive specialist. Mightn't he have been greater still were he to have expended less energy on the offensive end, where it apparently contributed, on average, nothing, and spent it instead on the defensive end, where he was very, very good indeed (average fake xRAPM of 4.8).

Now, maybe the apparent fake xRAPM penalty of +2 for top players falls disproportionately on the offensive side of the ledger, and the "real" picture is one of a more balanced player (in terms of productiveness). But, I'm not holding my breath on that.

I wonder where do voters rank Allen Iverson? I see votes for players that never got past the first round...

APBRmetrics: The discussion of the analysis of basketball through objective evidence, especially basketball statistics.