APBRmetrics

[MartyW]

I noticed there was a little debate a while ago about the greatest 10 players of all time so I just thought I throw in a few lists of the greats from books and the net, some ordered some not.
I just thought I'd throw it out there to discuss, not that I'm advocating (individual) metrics to solve whose best.
If someone wants to convert this to a viewable table that'd be great.

http://rapidshare.com/files/241408697/Basketball_50-100_Greatest.xls.html

I didn't include the ABA team or the Pre 50th Anniversary teams because of their datedness, or the TNT "Next 10" addendum to th 50th Anniversary, because well, I'd already uploaded the file.
I did include 2 rankings from one site (Whose best) because it seems to have changed some rankings for reasons I couldn't (be bothered to) find out why.

ed- Here's a link for The TNT "Next 10" and other teams just in case anyone cares http://hoopedia.nba.com/index.php?title=50_Greatest_Players

[Mike G]

Nobody else has done anything with this, so I have. I've also placed my own rankings alongside the others. In possibly chronological order, most-recently updated listed first, trying to give equal weight to each.

Key:
MG -- yours truly, thru 2008
IH -- Inside Hoops, date unknown
KT -- Keith Thompson, Heroes of the Hardcourt, 2005
EK -- Elliot Kalb, Who's Better, Who's Best in Basketball, 2003
WB? -- Who's Best, author and date unknown
WP -- Wayne Patterson and Lisa Fisher, 100 Greatest Basketball Players, 1988.
Unranked but included in this listing: The 1996 50th Anniversary Top 50; Alex Sachare, 100 Greatest Basketball Players of All Time, 1996; Assoc Press Player of the Century
(top 10), 1999; and APBRmetrics' Ring of Honour endeavor (a handful of participants) from a couple of years ago.

Top 50 according to these 10 listings:

[Mike G]

Here's a top 50 based only on the 5 lists with actual rankings, from within the last 10 years.

[Chicago76]

I have another list for you from Athlon in 1998. This was done more or less as a response to the NBA at 50 list:

1-Jordan
2-Russell
3-Jabbar
4-Robertson
5-Johnson
6-Bird
7-Chamberlain
8-Baylor
9-West
10-Pettit
11-Havlicek
12-Olajuwon
13-Cousy
14-Erving
15-Mikan
16-M. Malone
17-Schayes
18-Frazier
19-Pippen
20-K. Malone
21-Stockton
22-Barkley
23-Barry
24-Hayes
25-Daniels
26-Sharman
27-Cunningham
28-Cowens
29-Drexler
30-McGinnis
31-Robinson
32-Thomas
33-McAdoo
34-Gilmore
35-Bing
36-Archibald
37-Gervin
38-Hawkins
39-McHale
40-Reed
41-Haywood
42-Maravich
43-Ewing
44-Issel
45-Lucas
46-Greer
47-Rodman
48-Bellamy
49-Arizin
50-Thurmond


ESPN Daily Dime also put together a top 10 list of all time by position based upon a panel of former players, coaches, ESPN/ABC Media types. My google is acting up, but if you were to search Daily Dime all time centers ESPN, you would find it in a cache somewhere for centers. Replace centers with shooting guards for SGs, etc.

Americans (myself included) are pretty obsessive about rankings. We rate colleges, hospitals, cars, beers, children's strollers...anything. I personally like ranking players. It's a fun exercise/discussion, but it doesn't get much play here because it's pretty hard to do through simple (or complex) quantitative analysis across eras.

For example, if you were to use some minute-weighted PER with a replacement level of 12 or so, guys like Hal Greer, Dennis Johnson, or Joe Dumars wouldn't even break the top 200. It wouldn't shock me if any of those players were left off a top 100 list, but they should at least be among the next 20 or so looking in. Mike's list works as well as any that relies upon numbers, given the difficulties of comparing numbers across eras. Doesn't mean I agree with it all, but it makes as much sense to me as any other.

[Mike G]

As obsessed as some of us can be about finding 'what can the numbers tell us', others are obsessed with preserving the status quo, regarding players' status. The number of players who are 'surely' alltime top 10 is way more than 10.

How about the authors of list #6 in the first chart I posted? -- WP -- Patterson and Fisher, 1988. It's striking in that some players are ranked much higher in that list: Bird (4), West (7), Cousy (9), Barry (10), Jerry Lucas (20), Arizin (24), Walton (25), Maravich (11 <-- not a typo).

By my count, 53 of their 100 were white players. Even thru 1988, that's an overload; though perhaps in 1988, fewer would notice?

One thing that makes pure-numbers a better filter than such qualifiers as 'awards' and 'inclusion on previous lists' is that there's no built-in bias. I've found Hall of Fame bias against not only black players, but against tall and midwestern players. The average voter is probably shorter, whiter, and closer to the NY-LA geo-cultural axis than the average NBA/ABA star.
_________________
`
36% of all statistics are wrong

[ecumenopolis0]

[Chicago76]

ecumenopolis0,
The issues you mentioned are just a few of the reasons why normalizing stats across eras is so difficult.

It's funny you mentioned free throw %, because that is probably the most "pure" and directly comparable stat out there, however, even FT% for the same player would be subject to some adjustment. Lighting was worse, the rims less consistent, and it wouldn't surprise me if guys were shooting at 9'11" baskets one night and at 10'1" baskets the next.

First, there is the question of what you're trying to measure: how a player instantly transplanted from 1950 would do in 2009, or how a player from 1950, given the training, nutrition, and equipment available today would produce in 2009. The latter assumption is at least somewhat possible to approximate, because we can see a constant evolution in how people play as they bridge various eras and gain access to new innovations/developments: Russell played against Chamberlain, who played against Jabbar, who played against Parish, Malone, Olajuwon, etc.

Rules changes: how would West perform with a 3 pt line and Reggie Miller without? How much do you add to West's point production, and given the change in strategy, how much would you subtract from a big man West was playing with?

Stat interpretations: how many assists would Oscar Robertson accumulate with a more liberal interpretation of an assist? On average, you would think his numbers would be higher, but what if the Cincy scorekeeper was using a liberal definition of assist back in the day? Applying a league avg. ast/FGM ratio might overstate the adjustment materially.

Pace adjustment: we can normalize pace, but that is a purely linear adjustment that impacts players consistently. Changing the pace of play could actually positively impact some players' touches per possession, while negatively impacting others. For example, it is observable that the slower the pace, the higher the frequency of high usage players. It is a lot more common to find a 28% usage player today than it was in the early 80s, even controlling for expansion. Presumably, that's because teams take more time to get the ball to their key player. At a slower pace, your number three option might have a reduced role in the offense while your number one option might play a larger role. For example, if Moses Malone's Sixers played a Knicks-style pace from the mid 90s, his scoring might not drop as much as you'd expect from a linear assumption, because the Sixers would more deliberately go to him on the offensive end. On the other hand, I could see a player like Steve Nash thrive even more playing at 105 possessions/game.

Talent around a player: this one is just about impossible due to the usage/efficiency guesswork, but if Bill Russell and Wilt Chamberlain switched teams, what would happen to their production?

League strength: we can measure the production of the same pool of players year over year and measure this against some baseline to get some sense of how league quality changes over time. For instance, if the same 200 players produce more the following year, we can assume the league got weaker. The issues of how much weaker and what the baseline expectation should be is open to debate.

Position evolution: some of this ties into rules changes, but casting that aside for a second, some positions (notably shooting guard) probably weren't fully developed and exploited. This was partially due to the lack of the 3, but ignoring this, the numbers posted from players at the "2" weren't very impressive. SGs were either your #2 ball handler or your #1 ball handler who also happened to be your best outside scorer. Just looking at the 10 NBA seasons prior to competition from the ABA, very few guys cast as a shooting guard were getting estimated PERs of 17+. Using a pretty liberal definition of shooting guard, there were only 4 who actually managed to pull it off for three of those ten years: Jerry West (PG/SG), Richie Guerin (PG?), Sam Jones, and Hal Greer.

General playing style issues: It's quite possible that some of Chamberlain's physical advantages would diminish in the modern era as he was so far ahead of the curve in strength training vs. his contemporaries. It's also entirely possible that a center on a slower pace modern team like an Ilgauskas or Smits wouldn't have the wheels needed to play an early 60s pace.

Normalizing play across eras will generally get you closer to a "pure" estimate of a player's production. The issue is that so many assumptions are being normalized that the margin of error becomes pretty substantial.

[Mike G]

Excellent overview. Allow me to quibble just a bit --

[Chicago76]

I handpicked a couple of non-elite, but very good centers over the last 20 years who happened to have mobility and foot problems. For some players, the wear and tear under the basket in a slower paced game may have a greater impact. For a guy like Smits, who never really could play high minutes, and who also happened to have a lot of nerve damage in his feet, and who needed time to get his touches within an offense, the chuck it up pace of the early 60s would most likely be more detrimental than more physical play near the basket of the 90s.

Overall, I think your list makes sense. The things that don't (to me at least) may come down to difficult to quantify usage changes if we are to assume a constant pace. For example: Moses Malone vs. Patrick Ewing. Ewing was a player who dominated the ball in a slow paced offense that happened to take the time to feed him. Comparing their stats at the same age, Malone's usage was lower than Ewing's in all but a couple of their "prime" years. This was obviously going to be the case during Malone's years with the Sixers playing with Erving, but it still occured in 7 of the 9 times at ages 23-26 and 31-35 when Malone wasn't on the Sixers. Give Malone a 90 poss/game pace, his usage could realistically go up substantially, resulting in more pace-controlled pts, and his rating in your system may be more like 13-17. Likewise, play Ewing's prime years from 1977-1987, and his ranking might drop into the 30s. This is just a guess as to why Ewing is rated relatively higher on your list than most lists, and why Malone is rated relatively lower.

Maravich tends to be wildly overrated on most of these lists, and you have him at 200+. The truth may be somewhere in between. He was a hell of a shooter, but also the definition of a chucker who took a lot of bad shots. Put him on a slower paced team that runs the fast break opportunistically, put a bit of a leash on him, and he might come out looking much better. His short stint in Boston at the very end of his career saw a nice bump in his PER vs. the prior few years. Maravich and free wheeling go hand in hand, but I could see him flourish in a half court game 90 poss/game scenario where his ball handling and passing run the half court and his shooting is a little more selective.

This isn't a knock on your system. You're comparing thousands of players on teams with different personnel, playing with different rules, different prevailing playing league styles, different league strengths, etc. There is no clear adjustment for how a pace/personnel change would impact pace-adjusted stats.

Linear adjustments are the only way to realistically go about doing this. Most of the time, these assumptions will relatively wash out, but given the size of the player pool, there will be some cases where they won't.

[Mike G]

Ewing was a more versatile player than Moses. He was a beast defensively. If I change his rate of blocks to equal Moses', he's barely the lesser player.

Ewing anchored a premier NBA defense for a team that was perennially in contention. His minutes were 11.4% in playoffs, while Moses' were just 7.5% . If I switch those, their relative ranking also flips.

Moses did just 2 things very, very well. He was #1 at his position when other centers were in decline. Ewing's numbers were procured vs some alltime great opposing C's.

It's intriguing to speculate about a player's reaction to a different pace. There's actually quite a range of pace among teams in any given season, though; and we gather players' production within whatever system. A team that's built around a player should accomodate him. No one forces your team to play at a different pace, in 1952 or in 2009.

Maravich wasn't just a chucker, but had his greatest 'success' with bad teams, including college. High TO rates. Bad shooting%. The definition of 'overrated', if you ask me.
_________________
`
36% of all statistics are wrong