Man vs Machine: Human and Analytical Evaluations of NBA Greats
By: Ian Geertsen
Who are the greatest players of all time, and how are they ranked? These are timeless questions asked by every fan of every sport, and quite frankly are questions that I can not hope to answer. What I do hope, though, is that this piece adds a new perspective for how we look at the best players in the game of basketball by taking advantage of the advanced era we live in today.
For years, the ‘eye test’ was the only test; today, things are different. Never before has there been so much data readily available, and even if only a portion is available to the common man, it’s still enough to make your head spin. Through this analysis, I plan to compare these two lenses of viewing players—impression vs impact metric, man vs machine—and see how they rank the greatest players in NBA history.
Before diving into any metrics, I first wanted to take a look at who we, the people, view as the greatest players of all time. In order to get a better feel for how people view these players, I decided to split this into three subgroups: fans, media, and experts. To judge who these groups view as their NBA GOATs, I simply found as many polls, votes, and personal lists as possible. While this task sounds simple enough, finding enough of these sources to create even a measly sample size proved much harder than expected. Then, once I found enough lists for each subgroup, I assigned weights to each list based mainly on the sample sizes of the sources, as well as the overall quality, and used those weights to calculate an average ranking for each group.
Finding media sources was by far the easiest out of these three categories. I’ll be the first to admit that not all of these sources come from qualified analysts, but because I wanted to find an accurate view of how all of the media views these NBA greats, I actually see this as a positive. Despite the varying credibility of these sources—and knowledgeability of their authors—I did see relatively consistent results across these eight media sources; for each of these lists, for instance, the same 12 players found their way into the top fifteen one way or another. All in all, this is what the results from the media sources look like:
Finding fan sources, or credible fan sources, I should say, ended up being a lot harder than I thought it would be. In the end, I ended up incorporating a few polls with questionable methodology and incomplete results in order to give myself a larger sample size—this lack in quality is reflected in the weighting of these sources. I did my best to weigh these sources based on the number of responses they supplied, although this is inherently tied to the quality of their responses. As with the media sources, I am also pleased with the levels of expertise these sources have to offer (or the lack thereof). While there are a fair share of polls coming from discussion boards filled with dedicated fans, there are also votes cast from seemingly random websites which clearly have responses from more casual basketball viewers. Additionally, I see a healthy balance not only in the information gap of these fans but in the age gap as well. Some of these sources (i.e. the reddit polls) obviously come from a younger audience, while other websites are clearly aimed at targeting the viewership of older generations. Again, I believe that this wider range of ages and knowledge only contributes to a more well-rounded picture of how fans view the best players in NBA history. Also, some of the polls I included were not necessarily targeting how players are ranked but rather asking which player is the greatest of all time. While these sources are asking a different question, I found the data they provide was useful enough to be included in the analysis, although the lack of quality in these GOAT sources is noted in the weighting. Here’s how the fan results shook out:
Finding sources I could qualify as experts also proved to be more difficult than anticipated, mostly because I had such a hard time seeking out individual NBA experts’ lists. The only two such lists I managed to find came from analysts Ben Taylor and Bill Simmons. I’ll be the first to say that these two accomplished authors, podcast hosts, and analysts are about as good as they come as far as expert sources are concerned, but I would have loved to find more lists coming from analysts of similar stature. I will also note that some of the sources used in the expert panel have also been used in other groups, namely ESPN’s ranking of the top 74 players of all time, which could be seen as a source of expert opinions on its own. For this group of sources, the weighting I came up with was a bit arbitrary, although I believe it is the optimal system given the sources I had available. I wanted to place the most emphasis on ESPN’s top 74 players ranking, as this is a list compiled by multiple NBA experts, although I also wanted the individual opinions of Taylor and Simmons to have meaningful weight as well. Including NBA Math and Apbrmetrics adds a more analytical perspective to the conversation and, while not professionals, RealGM holds one of the most knowledgeable NBA fanbases on the internet; these sources are valuable in that they round out the overall ranking, although giving to much weight to any one of them would be inaccurate in the greater scheme of things.
Finally, we can compare these lists to one another, as well as see the equally-weighted average of all three rankings. While I have been including lists of up to 15 players, I felt that only going 12 down for these average lists was acceptable because these were the only 12 players who consistently found their way onto almost every top 15.
The Top Tier
Looking at these lists, there are a few things that stand out right away, the first of which being the top four names on every list. Jordan, LeBron, Kareem, and Magic managed to not only make the top four in all three of these lists, but to do so in the same order as well. This is not something that I expected when setting out to do this project; Jordan and Bron consistently at one and two I can understand, but having Kareem and Magic valued so highly by experts and fans alike speaks volumes to their legacies. These lists also shows the dominance and explains the reverence of one particular team in terms of all time greats; the Los Angeles Lakers. When looking at the aggregate of these three panels, we can see that four of the top five players wore a Lakers Jersey at some point in their career; when observing the top ten, we can see that six players donned the legendary purple and gold. The franchise that comes second in this measure is the Boston Celtics with three players in this top ten—including one abysmal year of an elderly Shaquille O’Neal. Now, this does make you wonder how much these players built the Lakers legacy or vice versa, and whether some of these players would be so high on this list had they not worn the purple and gold during their primes; this argument is applicable to the Celtics players as well, as these are the two most prestigious and winningest franchises in league history. The Celtics and Lakers rank first and second in NBA championships, winning 17 and 16 times respectively, and since the 1950s both teams have one at least one championship in every decade but two. While this success obviously comes from great players, the attention and fame that being on one of these teams brings often far outweighs that which small-market players receive, which can have a great effect on a player’s legacy. Take Kevin Garnett for example; despite dominating the league in the mid 2000s, he didn’t receive nearly as much credit for his performances as other players on bigger market teams, and it took winning a ring in Boston for him to finally earn his overdue recognition.
Russell and Chamberlain
Left or right. Stop or go. Russell or Chamberlain. A tale as old as time and twice as worn, I won’t be the first to look at these two Goliath’s careers, and I most certainly won’t be the last. On one hand we have Bill Russell, a 6’9” (at least) olympic-level high jumper whose defensive dominance has yet to be seen again. On the other, we have Wilt Chamberlain, the offensive dynamo who averaged a cool 30 PPG over his NBA career—including a season of averaging 50 points and 25 rebounds over 80 games—and famously scored the leagues only 100-point game. The debate between these two stalwarts of the game goes beyond individual accolades and into the fundamentals of the game itself. In my opinion, the fact that we saw these two players, who dominated the game in such different fashions, play in the same era is truly amazing. The 7’1” Wilt’s scoring and flashiness makes his game the fan favorite, despite the birth of rings Russell was able to earn in his career. But when taking a closer look at player efficiency, impact, and their meetings in the playoffs, it’s not surprising to me that Russell is deemed superior to Chamberlain by the experts in the field. In 49 playoff matchups throughout their careers, Wilt outscored Russell 25.7 to 14.9, and outrebounded him 28 to 24.7; while this at first appears to be a damning indictment for Russell’s case, it’s worth noting that this 25.7 points for Wilt was nearly five points below his regular season career points per game, and while he might have been outscored in the individual matchup, Russell consistently held Wilt to a lower efficiency in their matchup, in addition to winning most of their contests.
Bird, Duncan, Shaq
Considered one of the greatest small forwards, power forwards, and centers of all time respectively, it is impossible to ignore the mark these three left of the game. In Bird we got to see one of the smartest players the game has ever seen dominate despite his athletic limitations and career-shortening injuries. In Shaq, we saw a different kind of beast; someone who didn’t make his money off his IQ and skill, but rather one of the great physical specimens any sport has ever seen, who dominated by sheer brute force. Somewhere in the middle is Duncan; he had size and strength, he had smarts and skill, but lacked the large personalities the other two players have come to be known for. Unlike these two, Duncan spent his illustrious career playing for a small-market organization. Each of these three lists had these three players adjacent to one another, yet each list had them in a different order, as we can see here.
These results both are and are not surprising to me. It makes sense that the media would favor Shaq and his loud personality while lowering the more modest and quiet Duncan, just as it clicks that the experts would favor Bird’s legendary shooting, intelligence, and feel for the game over Shaq’s awesome physical superiority. At the same time, the fact that each group would have these three players together but in different orders is astounding to me, and speaks highly to each of their three unique yet commanding legacies.
If any player on this list deserved to be in a tier of their own, it’s Kobe. Put at 10th, 7th, and 11th by the media, fans, and experts respectively, it’s not very odd to think that fans would put him the highest while experts the lowest. He never was one to take the most efficient shots, but his fabled work ethic, uncompromising skill, and killer mentality will always make him not only a fan favorite—there’s a reason why we yell “Kobe” when shooting wads of paper at trash bins—but also one of the greatest players this game has ever offered us. Mentored by one of the greatest, he went on to teach and inspire the next generation of players, leaving the league better than he found it. I’m not here to talk about the on and off court greatness of Kobe Bryant, but to me, these lists go further in cementing Kobe not only in the greatest ever conversation, but in our hearts as well. Rest in peace.
Robertson and Olajuwon
These two men played two very different games in two very different eras, although they are both often brought up in this section of the greatest conversation. Olajuwon was one of the best defensive talents in league history and a star on the offensive end as well, but was hindered by a poor front office who didn’t put the right players around his generational talent. Robertson was once famous for amazingly averaging a triple-double over the course of an entire season. Pretty astounding accomplishment—or at least it was until Westbrook did it like eight times. These are two of the games most impactful players without a doubt, and I can’t wait till we see how these players are judged by the advanced metrics as well.
To make my life a little easier, I decided to create a short list consisting of a handful of players to look at from the lens of advanced impact metrics. That list will include the 12 players mentioned above, as well as a few editions which I will get into now.
I was very surprised at the lack of Robinson which greeted me while making these lists; I don’t think I’m alone when I say that I think he deserves a seat at this table. And, as the numbers will show, that seat may not be as far in the back as you might think.
Dr. J and The Logo
In addition to the players already mentioned, I wanted to add some of the elder greats who might have been glanced over. By including Jerry West we get to look at another offensive engine who played before the modern era, and in Erving we can look at an undeniable talent who has a reputation of being underrepresented by his statistics. Also, these players are legends in the minds and hearts of basketball followers that would be sorely missed from an all-time greats list.
Curry and KD
In the same way that I wanted to include the older West and Erving, I wanted to add some of the modern day greats to this list as well. To do this, I’ve enlisted the help of one of the best duo’s in the history of the league, Curry and Durant. While these estranged teammates probably don’t talk to each other much in real life, they are inarguably two of the best players in today’s league, especially on the offensive end; while I don’t expect them to end up near the top, I do think these two players will be firmly in the conversation for two of the best to ever do it by the time their careers are over.
As a counter to these lists, I wanted to look at these same elite players with the opposite approach; using data and advanced metrics. Before I get into how these metrics were used, though, I want to give you my reasoning for why I chose them. For this part of the analysis, I will be looking at these 10 metrics: career regular season VORP, Win shares, BPM, and PER; RAPTOR +/- and RAPTOR JAWS; PIPM +/- and PIPM Wins Added; and WOWYR and CORP.
Included in these metrics are some box score-based metrics like Box Plus/Minus (BPM) and Player Efficiency Rating (PER), which look at an individual’s box statistics to try and quantify a player’s performance. Value Over Replacement Player (VORP) uses BPM—a rate statistic—and estimates that players overall contribution to their team compared to a standard ‘replacement player.’ Similarly, Win Shares also attempt to estimate a player’s contribution to their team, this time by allocating ‘credit’ for a team’s wins based on box statistics. Next we have Nate Silver’s RAPTOR, a modern metric which combines box statistics and on-off data to determine a holistic view of a player—the box component is weighted more heavily in the metric’s calculations, however. In this study, I'll be looking at two types of RAPTOR: RAPTOR Plus-Minus and RAPTOR JAWS. Like BPM, the plus-minus component of RAPTOR measures the points players contribute to the offensive and defensive end per 100 possessions. In this way, the box component of RAPTOR is similar to BPM, although RAPTOR does include player tracking and pay-by-play data that is absent in BPM. The plus-minus statistics are used to calculate a player’s Wins Above Replacement (WAR) for that season, which can then be accumulated year-by-year; when averaging a player’s career WAR (including the postseason) with his peak seven years WAR we get a players RAPTOR JAWS (Jaffe WAR Score system), another metric I will be looking at in this study. Like RAPTOR, our next metric Player Impact Plus-Minus (PIPM) also includes a box component, although this metric combines the box statistics with luck-adjusted plus-minus data and a few interaction terms to estimate a player’s seasonal value. This luck-adjustment, designed by Nathan Walker, estimates a team’s offensive and defensive ratings while accounting for luck to try and reduce variance, and this component is really which separates this metric from other contemporary metrics. The per-100 possessions estimates provided by PIPM can be converted into a cumulative Wins Added metric which—similarly to VORP and RAPTOR WAR—takes playing time into account and compares players with an artificial replacement level. Finally, we have our last two metrics, both of which come from Backpick’s Ben Taylor and rely very little on traditional box stats. Championship Odds over Replacement Player (CORP) evaluates a player’s global offense and defense based on how much impact they have on different lineups, adjusts for health, and accumulates all of the player’s seasons. Finally, an adjustment for longevity based on era is made. This process is much more complex compared to that of older metrics like Win Shares, in part because of its use of a championship odds calculator which automatically balances longevity and peak play. This, along with its low reliance on traditional box metrics, makes it a great metric to balance more traditional options. We also have With Or Without You Regressed (WOWYR) A non-box metric which examines your team’s performance in games which you play vs games you do not play. Instead of simply looking at wins and losses, with-or-without-you uses the Simple Rating System (SRS) to determine a team’s win-pace with and without any given player—SRS takes into account point differentials and adjusts for strength of schedule, making it highly predictive of future performance. WOWYR is simply the regressed version of WOWY, and is the final metric I will be using in this study.
Essentially, all of these statistics (excluding CORP and WOWYR) can be narrowed down to one of two types: a rate-based plus/minus type metric (i.e. BPM, PIPM) or a career points-accrued metric (i.e. Win Shares, Raptor JAWS). Why is this important? Well we need a healthy balance between these two kinds of metrics because each emphasizes a different aspect of the game.
To further analyze this, let’s take a look at two specific players; Micheal Jordan and Kobe Bryant. While relatively similar in their playing style and winning pedigree, their differing career arcs help demonstrate the difference in these two types of metrics. This is because metrics which judge players by assigning them an accrued value over their careers give more weight to longevity, while rate metrics give more weight to consistency and longer peaks. Take Jordan for example; widely considered to be the greatest to ever do it, Mike ends up at only fifth in all-time regular season Win Shares because he played a relatively short 15 year career. Compare that to Karl Malone and John Stockton, numbers 4 and 6 on this list, who each played in the league for 19 years. Considering that those 15 years for Jordan include his broken foot season and return from baseball season, both of which he played less than 20 games, the fact that he’s at five is still pretty impressive. Essentially, in this kind of metric at least, his shorter career gives Jordan a distinct disadvantage. Still don’t believe me? Well when we look at Win Shares per 48, guess who’s the goat? MJ.
On the flip side, we have rate-based plus-minus statistics; these do not add additional value to your career each year, but instead average out that value; this puts players who had slow starts to their NBA careers or played at a significantly lower level in the later stages of their careers at a disadvantage. Since Jordan did neither of these, it's no surprise to see him atop the rankings for all-time Box Plus/Minus. Jordan’s ability to play at such a high level is a rare find, however—and was probably helped by his multiple hiatuses from the game—looking at other careers, like Kobe’s, tells a different story. During April of the 2012-2013 season, Bryant suffered a now infamous achilles injury. He only managed to play six games the next season and only had two more seasons after that; 2013 also happened to be the last year his Lakers would see a playoff appearance. While Kobe’s 20-year career gives him an advantage in accumulatory metrics over players who balled for a shorter period, injuries, age, and an unimpressive surrounding cast led his final years in the game to be far from his most efficient, giving him a disadvantage in many plus/minus type metrics.
I won’t argue against the fact that Kobe fares better in accrual metrics than rate ones—the data clearly shows this, as we will see soon—in fact, I actually view this as a positive. If we are talking about what makes a player great, their career’s longevity certainly plays a role. By the same token, are many years of mediocrity really worth much when compared to a few years of elite level play? I’m not trying to sit here and say that these ten statistics can combine to perfectly represent a player’s career, that would be silly. You could say that these numbers will end up placing too much emphasis on the regular season and not enough on the playoffs, and that these metrics will put too much value in longevity and not enough in peak play. I’m not going to say you're wrong either. But I do believe that looking at these players from an analytical lens as opposed to a human one can provide some pretty valuable insights, which I will look at now.
The Short List
Before getting into these metrics, it’s important to note that some of these players—Russell, Wilt, Robertson, and West specifically—played before the merger, so long before we have enough recorded data to calculate all these metrics. Because of this, these players will only be included in about half of these rankings. To try and still have these players included, I decided to rank the players by using a point system; points are allotted at different increments based on how many players are available for that metric, although the total amount of points given out is the same for every metric. Without further adieu, here are the player’s raw values for each of the ten metrics:
The player’s ranks for each metric:
The point system adjustments:
And the players final rankings (including standard deviation):
Jordan and LeBron
While it was close—only two pairs of players had closer margins based on my point values—Jordan did edge out LeBron in this one. Jordan bested Bron in six of the ten metrics I used, and pretty amazingly neither player fell outside the top four in any of the measures; this explains why these two players were also the most consistent out of the bunch. By my adjusted point rankings, Lebron was just 3.4 points behind Mike, although the next player in my rankings, Wilt Chamberlain, was a gaping 40.05 points behind The King. To me, Jordan and Lebron’s results in this ranking to not say as much about one compared to the other as it does about the two of them compared to the field. These guys are on a completely different level, and judging by this set of analysis, nobody else can compete.
Wilt ended up at the third spot on the list, although I’ll admit that the data coming from him and other players from before the merger should be taken a bit cautiously. Because of their small sample sizes, I expected these older figures’ rankings to be much more variant, as is measured by my standard deviation rankings. This was exactly the case for Robertson, West, and Russell, the other players to play before the modern era, as all of them ended up in the bottom five for standard deviation—meaning their deviations were very large. Not Wilt though. Despite having less than half the sample size of the other hoopers, Wilt finished with the third lowest standard deviation; a result which I would never have expected.
Yep, you read that right. Based on the aggregate of these metrics, David Robsinson shakes out to be the fourth player on this aggregated ranking. He did so with good consistency as well; ranking seventh in standard deviation, Robinson only finished within the top three on one list (PIPM +/-) and outside the top nine in one list (CORP). Right away when I began collecting this data, Robinson’s statistical dominance over players I would’ve considered much better than him astounded me. Obviously he has benefitted by playing in an excellent Spurs system, but to attribute this level of success to a coach and system alone would be a big mistake. Robinson did have a relatively short career for such a successful star, playing just 14 years, although this is partially due to the fact that he came in the league at 24 after serving in the US Navy. He got off to a hot start, winning ROTY in 1990, and in his 14 years he was an All Star and All NBA player ten times. This shorter career, combined with a sustained level of elite play, makes it unsurprising to see that Robinson dominated the plus-minus metrics, finishing 4th in BPM, 5th in Raptor +/-, and 1st in PIPM +/-. In comparison, he finished just 7th in Raptor Jaws and PIPM Wins Added.
Magic and Duncan
Coming in just behind Robinson, these two NBA legends tied (yes, TIED) for the fifth spot on this list. Ranking as the eighth lowest player by standard deviation, Duncan was a little more consistent than Magic, who finished as the eleventh most consistent of these NBA greats. While these players both had amazing careers, they played very different games stylistically, making their value at a glance hard to compare. To say that one is better than the other probably depends on what lens you see them through. Magic, for instance, won over Duncan in both of the RAPTOR indexes, while Duncan beat out Magic in each of the PIPM metrics.
Oscar Robertson ended up finishing at the seventh-best rank on this list, thanks in large part to his very high ranking from the WOWYR metric. This result does back up the widely held belief that he was the original “offensive quarterback,” or the first player to fit the mold of the ball-dominant heliocentric stars we see today. His combination of scoring efficiency, high volume, and excellent passing led Oscar’s teams to consistently have a top offense in the league. Oscar likely benefits from his small sample size in this case, and his consistency reflects this, as he is ranked 13th out of 17 in standard deviation. While his incomplete statistical resume makes it impossible to do a full analysis, these numbers do show how elite he really was. It makes you wonder how many championships he could’ve won if it wasn’t a certain big man in Boston.
Kareem, to me, is probably one of the most interesting players on this list, given the level of accuracy these lists portray him with. Unlike some of the older players who dominated in the 60s, most of Abdul-Jabbar’s career can be tracked by modern metrics, although unlike the players who would dominate the generation after his, the data we have on him is imperfect and incomplete. Starting in the 1969-1970 season, Kareem would go on to play 20 years; four of the metrics I am using, however, only begin using data during the 1973-1974 season, and the two RAPTOR metrics only use data starting at 1976-1977. At a glance, it’s easy to say that Abdul-Jabbar still played at least 14 seasons under most of these metrics, so what’s the big deal? While I won’t deny that his game became much more polished and efficient as his career went on, Kareem dominated the league in these age 22-25 seasons. Kareem’s four ‘lost’ years were likely his four best years in terms of raw box scores; these four years constituted four of his highest five scoring seasons, including the 71-72 season where he averaged an astounding 34.8 PPG. Although he averaged 11.2 rebounds per game over his career, Jabbar grabbed over 16 TRBs three times during the four year period. During this same four year span, Jabbar also made four All-Star teams, three All-NBA first teams, one All-NBA second team, two All-NBA defensive second teams, was a two-time scoring champion, rookie of the year, and won two MVPs. And I could go on. For us, what this results in is a player who scales out extremely well in some categories, and extremely poorly in others. And, unsurprisingly, leads to a player with the most variance in their rankings on this list: Kareem had the highest standard deviation of all these players by a large margin. While he currently stacks up at eighth on this ranking, if we only looked at the four metrics which evaluate him over the course of his entire career, being PER, Win Shares, WOWYR, and CORP, his score would come out to be 74.75, exactly tied with Wilt Chamberlain for third on this list. Maybe even more fascinating, out of these four metrics which completely track Abdul-Jabbar’s data, Kareem ranks first in two of them—that’s first not just out of players on this list, but out of all players. While I can’t say for certain where he should have ended up on this list, I am very comfortable saying that this position under-rates him. The question is how much so.
The second most inconsistently ranked player on this list, Bill Russell is another star to whom this exercise probably does not do justice. While he was able to make his way into the top five according to some metrics, Russell ended up dead last in PER out of these 17 players; in fact, PER really dislikes Russell’s game, ranking him at just 118th all time. I don’t know if John Hollinger has any personal grievances with Russell, but I’d warrant it’s worth looking into.
Ending up at 10th in ranking and 12th in consistency, can’t say there are any big surprises here. An interesting thing I noted about O’Neal was how certain types of metrics really like Shaq more than others, specifically accumulatory metrics, which value Shaq much higher than plus/minus ones tend to. This is something I’ll discuss in further detail later, but I think Shaq is a great example of archetype.
Olajuwon and Bird
Although very different, these two bigs ended up placing just 0.2 points away from each other in this exercise. Coming in at numbers 11 and 12, this is about where many people have Hakeem lined up, although Bird is valued lower on this list than many would expect. Part of this is because of his short career; Bird played just 12 full seasons, although I don’t actually think that this data is greatly misrepresenting him. In fact, Bird showed very low variance in his placements—he ranked as the fourth most consistent player on this list—showing that most metrics agree when it comes to evaluating Larry. Olajuwon finished in the top half of players in consistency as well, placing at number six in this category. As an interesting side note, it is surprising how closely these two greats were ranked in both ranking and variance when you consider that Bird placed in the top fifteen out of all NBA players for seven of these rankings while Olajuwon only did so for three of them.
Jerry West finished thirteenth in both his ranking and his standard deviation; this inconsistency isn’t that surprising when you consider that West placed 13th—16th on three of his available metrics and fifth in his last available one. West, like the other players who came before the modern era, does not have enough data to be evaluated on the same foot as his peers, although I believe that this exercise does confirm my belief that he should be placed somewhere near the back of this kind of list. This is especially true when you compare him to the other players who played in this same era.
While I would’ve loved to see Kobe end up higher on this list, this seems to be another unsurprising placement. Unlike most guards, Bryant was placed near the end of the list on plus/minus metrics like RAPTOR, BPM, and PIPM. Among the players on this short list, he actually was last on the list for BPM and PIPM, and placed second to last for RAPTOR. Career-accumulation metrics do have a higher valuation on the Mamba, though, which is why we can see him ranked 14th on the list with below-average consistency in his rankings.
Coming in at second to last on these rankings, Erving’s combination of poor scoring in the points system with his high consistency quotient shows that he likely belongs on one of the lower tiers of this list. Although he was never ranked last in any single metric, Dr. J failed to rank higher than seventh in any one metric, and was especially devalued by more advanced metrics like PIPM, RAPTOR, and CORP. In fact, Erving’s average rank for PER, VORP, BPM, and Win Shares came out to a respectable 8.75, although his average rank for the aforementioned advanced impact metrics came out to be 12.5. Very respectable, but not elite on this list of all time greats.
Curry and Durant
While the story may be different by the time their careers come to an end, as of now these two current stars have to take a backseat to the legends of yesteryear under these conditions. Finishing ranked at 15th and 17th respectively, these two players just don’t have long enough careers to compete with players of this caliber. Durant only has 11 full seasons to his name while Steph has just nine, as is reflected in the metrics. These two players were bullied in career-accumulation type metrics and ended up finishing dead last in VORP and Win Shares among the short list players. The former teammates also finished last in CORP and WOWYR, again a result which will probably not remain true once their careers are done. The duo did see success in other types of metrics, though; Durant finished sixth in both BPM and PER, while Curry was ranked fourth in both PIPM and RAPTOR +/-. The drastic difference between Curry’s valuation by +/- metrics compared to accumulatory metrics explains why he finished placing third to last in consistency, ending up with a higher standard deviation than players like West and Robertson who had less than half his sample size. Durant, however, ended up finishing with the eighth highest consistency, showing that his ranking is likely more representative of the totality of his rankings as opposed to the combined force of opposing outliers. Still, only time can tell how these two amazing players will end up on the stage of legends.
The Bigger Picture
In addition to comparing these 17 players with one another, I also wanted to take a more general look at the elite players in league history, but this time opening it up to every player who has ever been in the league. To do that, I instead looked at all time rankings out of all players for these same metrics. To avoid redundancy, lessen the impact of outliers, and focus these rankings on crediting extremely high levels of play, I made a new point system to measure these players in a new way: I assigned a player five points for making the top three in a given metric, four points for making the top 4-6, 3 for being in the top 7-9, etc. In addition to this new ranking system, I also took players’ average rankings for these ten impact metrics and used these averages in case of any ties—I also capped out players who ranked extremely low on a list at 30 to avoid the over-weighing of outliers.
Before I reveal the results of this new list, I want to talk about how the players we just discussed shake out with this new system. Because the two methods I used are fairly similar at the end of the day, it’s unsurprising to see that the players end up looking pretty similar on this list as they did on the previous one. Again, Jordan barely beats out LeBron—they actually tied on their point adjustments, but they both ended up lightyears ahead of the field. What I found to be incredible was that, for these ten impact metrics, Jordan’s average rank was 2.3 and LeBron’s 2.7 out of all eligible NBA players. These aren’t personal lists of fan votes, but impartial analytical rankings. Incredible. Wilt finished at third once again, with Magic, Duncan, and Robinson all coming in very closely after. While Robinson was put on top and Duncan and Magic tied on my previous list, this time Duncan edged the others out and Magic and Robinson were tied. Next up were Robertson and Kareem, same as last time, although this new list flips Russell and Bird, moving Bill from 9th to 12th and Bird from 12th to 9th among the short list players. In between the two Celtic legends are Shaq and Hakeem, who both stay put at rankings of 10th and 11th out of short list players. Following Russell, we have Curry, West, Kobe, Durant, and Erving. This list likes up-and-comers Curry and Durant a little more while valuing the older stars West and Erving a little less, but no big surprises here. Finally, let’s take a look at our last list of NBA greats:
It’s worth mentioning that once we get into the five or less points range I could have included many different players, so this list doesn’t place Erving as exactly the 22nd best player in history, for example. While the placement of the 17 players we looked at before stayed relatively consistent compared to one another, the things that stand out from this list are the influx of new players. Most notably, John Stockton and Chris Paul managed to sneak their way into the top five, with Kevin Garnett and Karl Malone making their way into the top fifteen. While I would definitely consider all four of these players to be underrated—players who fail to win rings often have underrated careers, and these four only share one championship between them—I still have a hard time believing this list, especially concerning Paul and Stockton. Basketball, like any other sport, is very hard to boil down to one metric, especially on the defensive end. Because of this, many of these metrics end up putting a lot of emphasis on steals and assists: both of which Paul and Stockton have racked up over their careers. Stockton is first in all time assists and steals, with Paul coming in at seventh and eighth respectively. Despite the fact that these two were probably given a disproportionate boost for their longevity as well, the rankings on this list are not flukes. Stockton placed in the top three on three of these lists, and CP3 did so on two; Kareem was the only other player not named LeBron or Jordan to reach the mark of two top three placements. If you can believe it, Jordan and LeBron each accomplished this eight times.
In addition to these sneaky and crafty guards, two of the games most dominant big men also ended up in the top fifteen on this list: Garnett and Karl Malone. Despite their impressive resumes, both these guys can be hurt by their reputations; Malone never could get over the colossal hump that is Micheal Jordan, and Garnett had to join a superteam to get his only ring. Still, though, both of these guys were dominant forces in the league that stand up to players from any era, as seen by the fact that both players placed within the top six for two metrics.
Up until this point I have been using analytical tools to examine some of the best players in league history. But halfway through working on this analysis, I had a bit of an epiphany: why not use the players to examine the metrics? In doing these analyses, there were two major patterns which stood out to me, being players performances in plus/minus categories vs accumulated-points metrics and RAPTOR vs PIPM. More specifically, I found certain archetypes of players seemed to excel in one of these categories and other types of players in the other, as the data below show.
When looking at these elites we can see that RAPTOR metrics tend to favor playmaking guards and wings, while PIPM puts a higher value on dominant bigs. This might not be news to people who know these metrics well, but regardless I think it’s an interesting topic to discuss. RAPTOR is reputed to put a higher valuation on steals and assists, but I still wouldn’t have expected the metric to weigh guards so heavily. For RAPTOR plus-minus the first conventional big to show up is David Robinson at number eight, and for RAPTOR JAWS the first appearance from a big is Duncan at number six. Compare that to PIPM, which has Robinson first in its plus/minus metric and Duncan at second, and the difference is clear to see. The only real exception to this pattern is Larry Bird, although his incredible passing and playmaking may help to explain this small blip.
It is interesting to note that guards and wings also seemed to be valued higher by plus/minus metrics and bigs by career-accumulation ones, with a few exceptions. Plus-minus metrics like BPM, RAPTOR, and PIPM actually favor David Robinson and Larry Bird, and career accrued metrics like Win Shares, JAWS, and VORP favor guards Kobe Bryant and John Stockton. One reason for these anomalies that immediately stands out is the longevity of these players. Bird only had twelve full seasons while Robinson had only thirteen, explaining why metrics which add value to you year after year would value your career significantly lower. Stockton and Kobe, however, both played for nearly 20 years. Additionally, rate-based plus-minus metrics seem to put a higher value on playmaking, as facilitating forwards Bird and Draymond Green are valued higher by plus-minus metrics. The only player I found that seemed truly neutral when faced with this comparison was Julius Erving. Dr. J actually ranked 13th among all players in VORP, BPM, and Win Shares; an above average rebounder and shot blocker for a non-big and an above average passer and playmaker for a non-guard, the metrics reflect the well-roundedness of this wing’s game which is so highly sought after today—and was even harder to find in the 80s.
To cap everything off, let’s take one last look at at all of the lists we’ve compiled:
Everyone seems to agree that Jordan and James should be at the top, but that’s about where most of the similarities end. The impact metrics seem to value Wilt and Robertson higher than most people do, although this may be a result of a scarcity of data; on the flip side, Bill Russell seems to be devalued by the metrics compared to people’s opinion of him. Kareem is also valued much lower according to the metrics, although that’s with a few of his key seasons missing from the picture. Kobe appears to be the most devalued by metrics compared to the public eye out of all modern players, with Bird coming in close behind. Magic also seems to be slightly devalued in this manner. On the other hand, we have Duncan; Tim is the only modern player in most people’s top ten to actually seem undervalued compared to the metrics. There were also quite a few players who did not get close to anybody’s top ten or even top fifteen despite being ranked very highly by the advanced metrics: Chris Paul, John Stockton, David Robinson, Kevin Garnett, and Karl Malone.
To wrap everything up, I just wanted to share what I feel are some of the most important takeaways I got from this analysis.
Jordan and LeBron appear to have played with a much higher degree of dominance than every other modern player on this list. While the debate between them rages on, they really do appear to be in a tier of their own.
Impact metrics appear to really like Duncan and Robinson, but it is unclear how much of this is due to their careers on excellent Spurs teams. Had they been in great situations like this, would we be talking about guys like Olajuwon and Garnett instead?
It’s very hard to draw legitimate comparisons between players from different eras using this data, but the metrics do seem to value Wilt higher than other players from the fifties and sixties. These metrics do not seem to love Bill Russell, but I think there are certain caveats to note here; Russell excelled on the defensive end, the side of the floor which has always been harder to quantify, especially given the metrics available for those older players. Had he been playing in the modern era and consistently leading the league in defensive metrics, how would exercises like this one evaluate him then?
Where do Stockton and Paul belong in the greatest of all time conversation? Do the metrics drastically overvalue their careers, or have we been drastically undervaluing them? If I asked you to think of an all time great NBA player, the image of a 6’1” pass-first point guard might not come to mind. Still, these are two of the best shot creators and playmakers in league history—according to Ben Taylor, Stockton comes in at fourth and Paul at third for the top playmakers in NBA history—and these results beg the question if we are devaluing the service these players provide, or if the metrics are overvaluing their skillsets.
While working on this analysis, the one thing that was constantly brought to mind was how I could better use analytics and modern data to compare and classify players. And concerning the greatest players in league history, how can I use these metrics to constitute greatness; more generally, how do we define greatness? How should factors like longevity vs peak performance come into play? Out of the ten metrics I used in this exercise, four of them—VORP, Win Shares, Wins Added, and RAPTOR JAWS—put lots of weight on longevity, a fact which can be seen in the high placement Stockton, for instance, and their relatively low placement of Bird. These stats also for the most part are concerning regular season numbers; if a player plays significantly worse in the postseason, again like Stockton, how should that skew them, and how should guys who significantly elevated in the playoffs, like Reggie Miller, be seen? How much room should we leave room for the games intangibles: the clutch performers and Mamba mentalities? At the end of the day, the answers to these questions are interpretable on the individual level, and I can’t foresee there being a single, universally accepted method for measuring championship equity—despite some statisticians' best efforts. While this analysis is certainly far from perfect, I hope it has contributed something to the conversation of greatness, the greatest conversation this game has to offer.
Sources: basketball-reference.com, bball-index.com, fivethirtyeight.com, backpicks.com, espn.com, bleacherreport.com, clutchpoints.com, cbssports.com, foxnews.com, complex.com, slamonline.com, ranker.com, reddit.com, topendsports.com, realgm.com, ranblingeveron.com, nbamath.com, spotify.com, apbr.org, wikipedia.org