My GOAT formula and results (Peak, Prime, Longevity and all time) - 22 players updated (last - Isiah Thomas)

Joao Saraiva · Post #1 » by **Joao Saraiva** » Sun Jun 26, 2016 9:58 pm

I've been developing a formula to determine how good was a player in each year in the NBA. And ultimately, his career value, peak value and prime value (his best 5 years, don't have to be consecutive).

The Formula is:

Spoiler:

Explaining why did I choose this formula and why it's only for players after 1980

Spoiler:

1. I chose per game values because PER 100 doesn't indicate the burden per game. If a guy plays 20 minutes per game and averages the same as a guy who is playing 38, then the guy who is playing 38 should be rewarded for it.

2. I chose per game stats instead of per 36 for the same reason of the above.

3. I chose per game stats over totals because it's easier for me to see the correlation between and achieve top values that make me weight the categories more or less the same, or with the importance I feel they have.

4. PER and WS/48 are stats I agree with the formula. They create some absurd results, but together I feel they're not bad. They hold some weight in the formula because I believe they give context to the raw stats. Usually PER and WS/48 benefit the best players, the way I see it. Didn't use BPM because I disagree with the formula, and VORP for the same reason.

5. TS% felt important. Efficiency matter with your scoring.

6. DBPM is included but with less weight. Believe me, it makes a diference in the end. I felt the formula needed some sort of defensive impact measure. I don't trust DBPM that much, but it makes a fine adjustment on low weight.

7. The coefficients were used because it's not the same playing 82 games or 20 games. It's not the same having one great playoff round of four great playoff rounds. I understand this brings some winning bias in the post season, and so does WS/48 on the formula. However, I feel like it doesn't weight that much.

8. Peak Value seems to hold less weight in the formula but it doesn't. It's the most repeated year, since it's value is included in Longevity Value and Prime Value. Prime holds a lot of weight, and I'd like to thank the guys who participated in a thread I created not a long time ago who said prime was the most important thing for them.

I could have gone with a 7 year prime, or 10 year prime. I opted for 5 because there are players who have 13 seasons, or 10 seasons I want to evaluate.

8. These are just numbers and they exclude any type of context. So, like any formula, it will give you results or weight things more or less than you agree, or that I agree. I don't think it's possible to evaluate a player with just a number. So take it as just an approach to what I believe rankings should be, not a defining value. Feel free to disagree also.

9. Using per game values ignores pace, so it's not good before 1980. It's also a bad idea before that because some stats are missing and the playoffs did not have the same amount of rounds.

10. I've made an accodales board that considers top 10 MVP shares, rings and finals MVPs. I tried not to give it a ton of weight since the MVP voting is based on others' opinions. It's subjective. However, it awards players who were top players in the league for more years (so it's a good addition for long prime players mostly). Rings and Finals MVPs were included to add winning value, since the objective of the game is to win.

I've only completed 22 players so far. I'll try to update as I add more players.

RESULTS

PEAK LIST
1. LeBron James 2009 - 314,08
2. Michael Jordan 1991 - 311,77
3. Shaquille O'Neal 2000 - 303,8
4. Tim Duncan 2003 - 294,39
5. Magic Johnson 1987 - 284,35
6. Hakeem Olajuwon 1994 - 283,07
7. Larry Bird 1986 - 280,23
8. Charles Barkley 1993 - 271,34
9. Dirk Nowitzki 2006 - 269,01
10. Dwight Howard 2009 - 262,98
11. Kevin Garnett 2004 - 261,87
12. Dwyane Wade 2006 - 259,18
13. Steph Curry 2016 - 258,36
14. Chris Paul 2008 - 258,24
15. Karl Malone 1998 - 255,41
16. David Robinson 1991 - 253,98
17. Kobe Bryant 2009 - 247,75
18. John Stockton 1989 - 236,59
19. Tracy McGrady 2003 - 235,23
20. Isiah Thomas 1985 - 222,53
21. Steve Nash 2005 - 219,86
22. Allen Iverson 2001 - 208,53

PRIME LIST
1. Michael Jordan 89, 90, 91, 92, 93 - 1466,04
2. LeBron James 09, 12, 13, 14, 16 - 1465
3. Shaquille O'Neal 95, 99, 00, 01, 02 - 1362,55
4. Hakeem Olajuwon 86, 88, 93, 94, 95 - 1347,77
5. Larry Bird 84, 85, 86, 87, 88 - 1334,32
6. Magic Johnson 82, 85, 87, 89, 91 - 1344,18
7. Tim Duncan 99, 02, 03, 05, 07 - 1317,54
8. Charles Barkley 89, 90, 91, 93, 96 - 1237,23
9. Karl Malone 92, 94, 96, 97, 98 - 1234,72
10. David Robinson 90, 91, 94, 96, 99 - 1226,45
11. Dirk Nowitzki 02, 03, 06, 09, 11 - 1197,90
12. Chris Paul 08, 11, 13, 15, 16 - 1187,96
13. Dwyane Wade 05, 06, 09, 10, 11 - 1184,42
14. Kevin Garnett 00, 02, 03, 04, 08 - 1182,83
15. Kobe Bryant 01, 02, 08, 09, 10 - 1175,05
16. Dwight Howard 08, 09, 10, 11, 14 - 1139,37
17. John Stockton 88, 89, 90, 92, 94 - 1102,19
18. Tracy McGrady 01, 02, 03, 05, 07 - 1052,38
19. Steve Nash - 03, 05, 06, 07, 10 - 1049,04
20. Isiah Thomas - 85, 86, 87, 88, 90 - 1037,8
21. Steph Curry - 11, 13, 14, 15, 16 - 1000,95
22. Allen Iverson 99, 01, 02, 03, 05 - 960,52

LONGEVITY LIST
1. Tim Duncan - 4113,07
2. Karl Malone - 3945,36
3. John Stockton - 3700,15
4. Shaquille O'Neal - 3678,21
5. Hakeem Olajuwon - 3555,37
6. Michael Jordan - 3437,75
7. Dirk Nowiztki - 3401,53
8. Kevin Garnett - 3316,78
9. Kobe Bryant - 3286,03
10. LeBron James - 3193,89
11. Charles Barkley - 3123,02
12. Magic Johnson - 3049,81
13. David Robinson - 2819,96
14. Larry Bird - 2765,95
15. Steve Nash - 2382,52
16. Dwyane Wade - 2239,95
17. Dwight Howard - 2088,45
18. Chris Paul - 2049,61
19. Isiah Thomas - 1985,92
20. Tracy McGrady - 1869,53
21. Allen Iverson - 1841,62
22. Steph Curry - 1128,89

ALL TIME LIST
1. Michael Jordan - 373,31
2. LeBron James - 353,30
3. Tim Duncan , 350,74
4. Shaquille O'Neal 345,87
5. Hakeem Olajuwon - 325,60
6. Magic Johnson - 323,02
7. Karl Malone - 320,69
8. Larry Bird - 313,43
9. Kobe Bryant - 303,28
10. Dirk Nowitzki - 289,67
11. Charles Barkley - 287,65
12. Kevin Garnett - 281, 27
13. David Robinson - 279,74
14. John Stockton - 263,85
15. Dwyane Wade - 252,29
16. Chris Paul - 249,1
17. Dwight Howard - 239,83
18. Steve Nash - 232,32
19. Tracy McGrady - 219,47
20. Isiah Thomas - 213,02
21. Allen Iverson - 205,76
22. Steph Curry - 194,59

What I'd like you to address

Spoiler:

The original post when created this thread that was united with another one I've made was the following

Spoiler:

Well can't help but feel it's not the same topic or anything, but if I can't have both threads active I'd rather have a thread to discuss the list and it's results.

Post #2 » by **Texas Chuck** » Sun Jun 26, 2016 10:13 pm

I value every positive value season in a player's career. I don't see them as just padding totals. Those seasons benefit actual teams and so they count. Kareem in the 80's, Duncan in the 10's---these were very valuable seasons. Those players should be credited for their performance in them.

I tend to put a lot less weight on a single peak year. In fact when trying to determine how good a player was at their absolute best I require at minimum a 3-year span. If one season stands way out like say 03 TMac, I average it down with surrounding seasons because I believe that gives me a truer picture of the player.

So when comparing players I like to look at best 3 year spans, best 5 year spans, best 10 year spans, and total career value.

BasketballFan7 · Post #3 » by **BasketballFan7** » Sun Jun 26, 2016 10:16 pm

I mostly judge on prime, but ideally I am getting a solid 7 or 10 year prime. On here prime is generally 5 years. So I guess that is some of the longetivity for me. Otherwise longetivity doesn't do much for me. The gap between a 9.0 player and a 9.5 player is so, so much more vital to a championship than the a 5.0 to a 6.0, for instance. What I mean is that I am looking for seasons that truly move the needle and that, talking about all time guys, low AS impact or low minutes really don't sway me. I'd much rather just go with the superior prime, even if the difference is incremental.

Peak to me is generally included in prime as well, because guys who have a singular season that stands out from their prime significantly generally don't impress me. However, peak would be of more importance if there are multiple seasons of that peak impact, or as a tiebreaker.

Idk, tough to do with percentages, but something like 70% prime, 20% peak, 10% longetivity.

JordansBulls · Post #4 » by **JordansBulls** » Sun Jun 26, 2016 10:19 pm

They are all important, but longevity matters if the player is still a top player in the league. Like Malone winning league mvp at 35 or MJ. But Shaq from 2007 to 2011 doesn't do much for him for example

Joao Saraiva · Post #5 » by **Joao Saraiva** » Sun Jun 26, 2016 10:24 pm

JordansBulls wrote:They are all important, but longevity matters if the player is still a top player in the league. Like Malone winning league mvp at 35 or MJ. But Shaq from 2007 to 2011 doesn't do much for him for example

Still, of course each season has it's value.

However, how much would you weight each category?

JordansBulls · Post #6 » by **JordansBulls** » Sun Jun 26, 2016 10:28 pm

Joao Saraiva wrote:
JordansBulls wrote:They are all important, but longevity matters if the player is still a top player in the league. Like Malone winning league mvp at 35 or MJ. But Shaq from 2007 to 2011 doesn't do much for him for example

Still, of course each season has it's value.

However, how much would you weight each category?

Probably something like:

Prime 45%
Peak 40%
Longevity 15%

Mainly because what you do in your prime and peak is more important than your latter years when you are fading unless of of course you are still the top player or two in the league.

mischievous · Post #7 » by **mischievous** » Sun Jun 26, 2016 11:04 pm

Can I use a 4th being accomplishments/accolades?

If i am to use the 3 you provided i'd say probably say something like 50 % prime, 30% peak and 20% longevity. Although when you think about it the 3 are kind of tied together, for example a peak is part of a prime so how exaxtly do they get seperate categories? Also, prime is a part of longevity, so that's why there isn't and shouldn't be some magical set standard to rank players.

Look at how good they were, for how long and what did they achieve in that span.

AceofSpades69 · Post #8 » by **AceofSpades69** » Sun Jun 26, 2016 11:12 pm

Generally speakin:

Prime : 45%
Peak: 30%
Titles (with a substantial contribute): 15%
Longevity: 10%

Of course this all means zero when one player is Penny Hardaway and the other is Chauncey Billups, but generally speaking, I guess that's my way of looking at a player's career.

Joao Saraiva · Post #9 » by **Joao Saraiva** » Sun Jun 26, 2016 11:29 pm

Wow. Too many people stating longevity under 20%. I'd expect peak (that is only one year) to be less weighted than longevity, because it is a much larger sample.

Joao Saraiva · Post #10 » by **Joao Saraiva** » Sun Jun 26, 2016 11:43 pm

mischievous wrote:Can I use a 4th being accomplishments/accolades?

If i am to use the 3 you provided i'd say probably say something like 50 % prime, 30% peak and 20% longevity. Although when you think about it the 3 are kind of tied together, for example a peak is part of a prime so how exaxtly do they get seperate categories? Also, prime is a part of longevity, so that's why there isn't and shouldn't be some magical set standard to rank players.

Look at how good they were, for how long and what did they achieve in that span.

I'm developing a formula to be honest, so I create these type of threads to help.

So far, the most difficulties I've had are:
- Ranking only players after 1980. The missed information from previous seasons makes me have to adjust a ton on the formula;
- Deciding how much I should weight longevity, peak and prime;
- Not much defensive information. Sure there are steals, DWS, BLK, DBPM but I really don't trust the numbers on D - they seem highly team oriented or a bit irrelevant. Getting a ton of steals more than the other guy doesn't mean you're a better defender;
- Coefficients that make the stats I'm choosing relevant enough.

I've only seen how the formula works on a few players (Duncan, MJ, LeBron, Kobe, Bird and Magic). Quite honestly, I'm happy with the results. Peak seems right, prime too, longevity too. (And respective rankings). The final career value I've achieved seems fine by me, with Duncan leading it by a tiny margin. So maybe I'm weighting longevity too much. I have made it so far with 0,3 * longevity value + 0,6 prime value + 0,1 * peak value. The peak value is also included on prime value and longevity value (well, it's a season among the others). So I felt it had enough weight. Prime is inside longevity too. I just feel like one year shouldn't get more weight, but the general opinion here doesn't agree. Thank you, and btw thank you all for your thoughts.

2klegend · Post #11 » by **2klegend** » Mon Jun 27, 2016 12:46 am

Realistically, 30%peak, 30% prime, 30% accolade, 10% longevity.

Peak is important. It basically show which player plays at their best and you can then project that if everything is equal, luck and injury free, we can expect similar career path result.

Prime is important. It shows how well that player sustains his level of play. For me, 5 years is the "break point" that show what that player can do.

Accolade is important. It shows that within stat-padding mindset, you have to win and show that you are a winner and not just stat-padding for the heck stat-padding (ala the Wilt of the universe). Of these 30% accolade weigh, I break off into percentage...
MVP 10%
Ring Main Focus 10%
---Complentary focus 5%
Final MVP 5%

Longevity is the least important. It doesn't show anything much except he decided to stay to play competitive. There are many reasons for that, money, love of the game, or lifestyle. Whatever it is, it is the least important but should encompass some value. Longevity is also the hardest to figure out how to reward it. I have trouble developing a formula for this but I'm coming very close.

sixeradelphia · Post #12 » by **sixeradelphia** » Mon Jun 27, 2016 1:00 am

Joao Saraiva wrote:Wow. Too many people stating longevity under 20%. I'd expect peak (that is only one year) to be less weighted than longevity, because it is a much larger sample.

I agree completely. Mine goes as follows

Prime 50%
Longevity 30%
Peak 20%

I personally find prime, longevity a much more impressive feat if able to contain it and improve on. Getting it all together for one year (see TMac) but then falling back to Earth the following 5 years makes the peak a wash in my eyes.

For example we have seen so many different versions of LeBron that I think his prime will become unprecedented and even though 2016 is not a "peak" year from a analytic view I believe it could be one of his best years in his career.

That's why I rate Duncan higher than Shaq overall, because even though Shaq was clearly a better peak player I can't ignore Duncan's longevity and continued success as a player. (Not only talking about rings either)

Not saying this is the right way of doing things just my .02

LakersLegacy · Post #13 » by **LakersLegacy** » Mon Jun 27, 2016 1:12 am

I like the idea of looking at it in terms of 3 year, 5 year, 7 year and 10 year time spans. I think you also have to make adjustments for injuries. I think some players are over/under rated because of 1 outlining year.

Joao Saraiva · Post #14 » by **Joao Saraiva** » Mon Jun 27, 2016 2:07 am

Once again thanks guys. A bit off topic but if anyone can reply, I feel rebounds are a bit overvalued in my formula vs assists.

If rebounds are valued as 1, how much do you think an assist should be rewarded?

I was thinking 1.8*TRB + 2*AST. It's just that grabbing 15 rebs seems a lot more likely than making 15 assists.

2klegend · Post #15 » by **2klegend** » Mon Jun 27, 2016 6:25 am

Joao Saraiva wrote:Once again thanks guys. A bit off topic but if anyone can reply, I feel rebounds are a bit overvalued in my formula vs assists.

If rebounds are valued as 1, how much do you think an assist should be rewarded?

I was thinking 1.8*TRB + 2*AST. It's just that grabbing 15 rebs seems a lot more likely than making 15 assists.

Go with Reb Rate and Ast Rate. Using raw data is misleading.

Blackmill · Post #16 » by **Blackmill** » Mon Jun 27, 2016 8:05 am

Joao Saraiva wrote:Once again thanks guys. A bit off topic but if anyone can reply, I feel rebounds are a bit overvalued in my formula vs assists.

If rebounds are valued as 1, how much do you think an assist should be rewarded?

I was thinking 1.8*TRB + 2*AST. It's just that grabbing 15 rebs seems a lot more likely than making 15 assists.

For starters, I think you may want to split up offensive rebounding and defensive rebounding. Although the two are equal from a team perspective, an individual can have greater relative influence on offensive rebounding than defensive rebounding, and the difference I speak of is large. Specifically, depending on which study you refer to, a 1% increase in individual offensive (defensive) rebounding probability increases the team's or lineup's rebounding probability by 0.4-0.6% (0.1-0.2%).

I can also say, having taken to watching games and recording every relevant action, that offensive rebounds are almost always contested and without another teammate in rebounding position, where else defensive rebounds are frequently uncontested and with another teammate in defensive position. This makes me think the coefficient for offensive rebounding ought to be on the higher end, at least for centers, who are responsible for most of the offensive rebounds I've recorded.

Regarding assists, the simple fact is a constant coefficient will relatively underrate good playmakers and overrate bad ones. Some function of raw FG%, APG, and maybe RPG could make for a decent coefficient. The raw FG% and RPG would help capture how close to the basket the player is operating. This is relevant to playmaking since drawing doubles and collapsing the defense, which occurs more frequently on post ups and penetration, lead to more open shots after the pass. The APG would try to capture the player's passing ability itself since better passers are more likely to gravitate towards playmaking.

That said, I think creating a good coefficient for assists is entirely impossible. The only work around is to determine the correct coefficient for each player individually by recording the types of passes they make. Assists are feasibly the most prone box score to differing wildly in terms of average value between various players.

Post #17 » by **tsherkin** » Mon Jun 27, 2016 11:19 am

I value prime longevity more than overall longevity. I factor that in somewhat, but it matters less to me. I don't really care, for example, what John Stockton was doing after after 98 too much. It's nice, it's valuable to a franchise if he's taking a step back on his pay (under modern salary rules, since obviously it wasn't a huge deal in his actual career) and everything, but older stars well past their prime aren't usually THAT useful except for the purposes of nostalgia. I mean, I love what Dirk is doing with Dallas, don't get me wrong, but these years aren't things I mostly consider when I'm examining him relative to his in-era and cross-era peers, you know?

But length of prime matters a lot. Peak matters, but there's got to be a minimum threshold. One-year peak doesn't mean as much to mean, or I'd weight someone like Bill Walton a little higher (although he takes a health penalty even in 76, 77). I dunno. At some point, holistic examination of the player becomes necessary because past a given threshold, most of these guys are damned good.

I'd say for me, prime weights like 70%, peak like 20% and longevity (if that means full breadth of career as opposed to prime longevity in particular) like 10%.

mischievous · Post #18 » by **mischievous** » Mon Jun 27, 2016 1:37 pm

tsherkin wrote:I value prime longevity more than overall longevity. I factor that in somewhat, but it matters less to me. I don't really care, for example, what John Stockton was doing after after 98 too much. It's nice, it's valuable to a franchise if he's taking a step back on his pay (under modern salary rules, since obviously it wasn't a huge deal in his actual career) and everything, but older stars well past their prime aren't usually THAT useful except for the purposes of nostalgia. I mean, I love what Dirk is doing with Dallas, don't get me wrong, but these years aren't things I mostly consider when I'm examining him relative to his in-era and cross-era peers, you know?

But length of prime matters a lot. Peak matters, but there's got to be a minimum threshold. One-year peak doesn't mean as much to mean, or I'd weight someone like Bill Walton a little higher (although he takes a health penalty even in 76, 77). I dunno. At some point, holistic examination of the player becomes necessary because past a given threshold, most of these guys are damned good.

I'd say for me, prime weights like 70%, peak like 20% and longevity (if that means full breadth of career as opposed to prime longevity in particular) like 10%.

Out of curiosity how much weight do you put on Duncan's post prime seasons like 2012-2015 for example? The Spurs made 2 finals, and 3 Wcf with him being still an all star player and defensive anchor. He even made all first team in 2013.

Post #19 » by **tsherkin** » Mon Jun 27, 2016 1:48 pm

mischievous wrote:Out of curiosity how much weight do you put on Duncan's post prime seasons like 2012-2015 for example? The Spurs made 2 finals, and 3 Wcf with him being still an all star player and defensive anchor. He even made all first team in 2013.

Yes. Those were high-end defensive seasons with reasonable offensive contribution. He's a puzzler. The absolute extremes are of interest and I'm not 100% sure how to handle those, given the balance of system, minutes, teammates and so forth. I'm still in the process of sorting through all of that, because it's a good question. Right at the moment, I have Duncan as a top 10 player, and I'm not really sure exactly where therein to rank him. My top 10 is pretty fluid these days.

Superbasketball · Post #20 » by **Superbasketball** » Mon Jun 27, 2016 2:04 pm

Doesn't longevity favor Lebron because he didn't play in college? Also if you look at Jordan he could have been the top player in the league for 3-4 more years if he didn't play baseball and retire in 98.