Based on some discussion that @cobainwasmurdered and I had in chat a few nights ago:
Trying to determine an "unbiased" method of measuring greatness in the NBA is always something I've been interested in. For years, I've played around with different ways of ranking based on measurable sets of data. Of course, I've seen the majority of players since the very early 90s play but I can't remember everyone and I certainly watched more of some players than others. As a result, trying to produce a list without using measurables or data is going to result in a "biased" list, which is exactly what I'm trying to avoid.
The easy part is narrowing down the four categories of data. All contribute equally to the final ranking, but -- depending on the player -- some categories may contribute more for one player than another. Also, I like tinkering with ratings by decade as styles of play are often similar over a ten year period and it allows for more recognition of players who may get lost in the shuffle in a ranking across all decades. Also, I decided to keep it post-merger -- there is just too much data missing from before that time.
I also looked at the NBA's All-Decade Teams released during this year's 75th anniversary and looked at who made the cut for each decade (note: whatever year the Finals occurred in was the decade I assigned for the entire year, e.g., 1999-2000 falls into the 2000s). I then looked at the chosen data (see below) for each player selected and weighted more heavily those that seemed to influence placement more. Obviously, an MVP is gonna show up on every list so that piece of data weighs heavily. Also, things like total points are going to weigh more heavily than the rest of the statistical categories because that's what voters (not only in the all-decade lists but in HOF selections) value more. The difference is subtle, but enough to influence placement in a cat. Bored yet?
Anyway, back to the four sets of data:
1. Statistical output: The easiest data to input as it's all raw numbers. I decided to go with the "big three" (PTS/REB/AST), PER (for the metric nerds), and annual leaderboard placement to reward guys who may have done really well in a cat for a couple of years but their total decade's numbers don't show it. This (and PER) helped guys who played for less than a full decade to pull a little closer to those who played the entirety.
2. Awards: This is essentially the "eye test" portion. Obviously, writers and voters are biased, but it at least gives me its own set of measurable data that I can use for ranking purposes. MVP obviously weighed the highest. I also considered MVP runners-up, All-NBA selections (more weight placed on 1st than 2nd or 3rd), DPOY, All-Defensive selections, and All-Star selections.
3. "Peak" value of a player: How good was this player in their best year? This was the hardest to measure, but I was able to come up with a combination of max-win shares combined with max-award voter shares to reach a suitable number. I considered adding in max-PER, but the standard deviation among the top players was small enough that I'm not sure it would make much of a difference, other than widen the gap between offensive-minded players (which PER favors) and guys like Rodman and Ben Wallace.
4. Postseason performance: Combination of cumulative statistical output that was multiplied based on how far a player got in the playoffs, as well as what they contributed to the squad. I only gave "bonuses" for reaching the conference finals or higher. I also gave a boost in this cat for Finals MVP, as I thought it more appropriate to use that here than in the awards section.
Now, with all of the egghead shit out of the way, let's look at the sample decade I used: the 2000s (I feel like 90s are done to death). For now, only players who made the 2000s all-decade team linked above are going to be listed since that's who I used to establish the "standards" that the remaining players would be measured against. I'll probably go back and add to this list when I do some more notables:
15. Amar'e Stoudemire
14. Paul Pierce
13. Ben Wallace
12. Tracy McGrady
11. Chauncey Billups
10. Dwyane Wade
9. Jason Kidd
8. Steve Nash
7. Allen Iverson
6. Dirk Nowitzki
5. LeBron James
4. Kevin Garnett
3. Shaquille O'Neal
2. Kobe Bryant
1. Tim Duncan
Number 1s in each cat (interesting in that a different name appeared for each one):
Stats: KG
Awards: Duncan
Peak: LeBron
Playoffs: Kobe
Number 15s in each cat:
Stats: Ben Wallace
Awards: Paul Pierce
Peak: Paul Pierce
Playoffs: Stoudemire (barely below McGrady, whose score in this cat killed his final ranking)
I'll be updating this to post some more names for the aughts and other decades.
Trying to determine an "unbiased" method of measuring greatness in the NBA is always something I've been interested in. For years, I've played around with different ways of ranking based on measurable sets of data. Of course, I've seen the majority of players since the very early 90s play but I can't remember everyone and I certainly watched more of some players than others. As a result, trying to produce a list without using measurables or data is going to result in a "biased" list, which is exactly what I'm trying to avoid.
The easy part is narrowing down the four categories of data. All contribute equally to the final ranking, but -- depending on the player -- some categories may contribute more for one player than another. Also, I like tinkering with ratings by decade as styles of play are often similar over a ten year period and it allows for more recognition of players who may get lost in the shuffle in a ranking across all decades. Also, I decided to keep it post-merger -- there is just too much data missing from before that time.
I also looked at the NBA's All-Decade Teams released during this year's 75th anniversary and looked at who made the cut for each decade (note: whatever year the Finals occurred in was the decade I assigned for the entire year, e.g., 1999-2000 falls into the 2000s). I then looked at the chosen data (see below) for each player selected and weighted more heavily those that seemed to influence placement more. Obviously, an MVP is gonna show up on every list so that piece of data weighs heavily. Also, things like total points are going to weigh more heavily than the rest of the statistical categories because that's what voters (not only in the all-decade lists but in HOF selections) value more. The difference is subtle, but enough to influence placement in a cat. Bored yet?
Anyway, back to the four sets of data:
1. Statistical output: The easiest data to input as it's all raw numbers. I decided to go with the "big three" (PTS/REB/AST), PER (for the metric nerds), and annual leaderboard placement to reward guys who may have done really well in a cat for a couple of years but their total decade's numbers don't show it. This (and PER) helped guys who played for less than a full decade to pull a little closer to those who played the entirety.
2. Awards: This is essentially the "eye test" portion. Obviously, writers and voters are biased, but it at least gives me its own set of measurable data that I can use for ranking purposes. MVP obviously weighed the highest. I also considered MVP runners-up, All-NBA selections (more weight placed on 1st than 2nd or 3rd), DPOY, All-Defensive selections, and All-Star selections.
3. "Peak" value of a player: How good was this player in their best year? This was the hardest to measure, but I was able to come up with a combination of max-win shares combined with max-award voter shares to reach a suitable number. I considered adding in max-PER, but the standard deviation among the top players was small enough that I'm not sure it would make much of a difference, other than widen the gap between offensive-minded players (which PER favors) and guys like Rodman and Ben Wallace.
4. Postseason performance: Combination of cumulative statistical output that was multiplied based on how far a player got in the playoffs, as well as what they contributed to the squad. I only gave "bonuses" for reaching the conference finals or higher. I also gave a boost in this cat for Finals MVP, as I thought it more appropriate to use that here than in the awards section.
Now, with all of the egghead shit out of the way, let's look at the sample decade I used: the 2000s (I feel like 90s are done to death). For now, only players who made the 2000s all-decade team linked above are going to be listed since that's who I used to establish the "standards" that the remaining players would be measured against. I'll probably go back and add to this list when I do some more notables:
15. Amar'e Stoudemire
14. Paul Pierce
13. Ben Wallace
12. Tracy McGrady
11. Chauncey Billups
10. Dwyane Wade
9. Jason Kidd
8. Steve Nash
7. Allen Iverson
6. Dirk Nowitzki
5. LeBron James
4. Kevin Garnett
3. Shaquille O'Neal
2. Kobe Bryant
1. Tim Duncan
Number 1s in each cat (interesting in that a different name appeared for each one):
Stats: KG
Awards: Duncan
Peak: LeBron
Playoffs: Kobe
Number 15s in each cat:
Stats: Ben Wallace
Awards: Paul Pierce
Peak: Paul Pierce
Playoffs: Stoudemire (barely below McGrady, whose score in this cat killed his final ranking)
I'll be updating this to post some more names for the aughts and other decades.
