My Personal NBA Top 50: How to build the best
My Personal NBA Top 50 is an interactive algorithm that generates a ranking of the greatest NBA careers of all time based on user-defined criteria. Which do you value more — career longevity or peak performance? Scoring volume or efficiency? Individual stats or team success? Tell us what matters most to you and we’ll create a custom Top-50 list based on your own value system.
When I was a kid, I had a Top-50 list for movies. It was expansive — a diverse collection of films spanning the range from early 1980s action comedies about sports to late 1980s funny sports movies with action scenes. There were probably (definitely) several sequels honored. I wrote the list on index cards and when I was bored — stuck in a backseat or in a waiting room — I would thumb through and re-evaluate my rankings. Titles were sometimes crossed out and replaced by a new release. Old favorites climbed and fell thru the rankings based on the cable TV offerings from that week. The list wasn’t immutable and it wasn’t universal. It was intensely personal- – a reflection of my own tastes and values, shifting constantly with my worldview.
And, of course, every Top-50 list is like that. You make a series of conscious (or subconscious) value judgments about the elements of your list, mash those criteria together using some unwritten formula stored deep inside the big mushy computer in your head, and then share your list with the world in one very definitive-sounding tweet. If your methodology is well thought out and your arguments persuasive, your rankings may resonate with others; maybe you’ll even get to write a book about your list. If not, you can always chalk it up to the Eye Test and remain secure in your absolute correctness. But, even if you fail to articulate your value system to others or fail to recognize it yourself — on some level, you’ve got a system.
My Personal NBA Top 50 is a project that’s meant to encourage us to confront our own value systems and to recognize the validity of ranking criteria distinct from our own. We may find biases lurking on our lists or deeper truths about the value of overlooked players. Or, if not, we will at least find Jeff Ruland. Because I know he will be in there, for sure.
Here’s how it works. You’ll be asked to input your opinion on the relative value of seven player traits: Superiority, Longevity, Efficiency, Team Success, Peak Form, Buckets and #Rangzzzz. The relative importance you give to these traits helps to define your personal ranking formula by assigning weights to a series of related stats (via Basketball-Reference, current through the 2017-18 season). For example, move the Superiority slider up to “important” and your rankings will become more heavily influenced by the number of All-NBA team selections and MVPs that each player earned during his career. Bang the #Rangzzzz button and your Top-50 will have a greater emphasis on the players who tallied the most championships. Swipe right on Buckets and guys with higher career scoring averages will climb to the top of the list.
We used three forms of Win Shares, a popular all-in-one statistic, to help define Longevity (career regular-season WS), Peak Form (best single-season WS), and Team Success (career playoff WS) and we defined a player’s efficiency by his career true-shooting percentage. Anybody who was an All-Star at least once had a chance to be ranked (417 candidates, in all).
After inputting your basic ranking criteria you can use a series of multipliers to refine your rankings based on era, position, or team. The levers for decade, allow you to discount stats from, say, the 1950s relative to today or to define your own personal Golden Era of basketball. Similarly, you can decide whether you want to give players credit for their contributions in the ABA. The position levers can also push players up or down your list based on their designation as a guard, forward, or center. Just for kicks, you can also give players from your favorite team a boost or sandbag players from your most-hated rival (players were associated with any team for which they played at least 10,000 minutes).
Basically, the algorithm is like your own personal G.O.A.T. factory — you man the controls and we’ll crank the gears and distill your settings into a tasty Top-50 list.
The tool’s not perfect, but it is pretty darn flexible. And to demonstrate a little of what it can do — to pressure-test the system, so to speak — we’re going to take a stab at reverse-engineering three expert-defined Top-50 lists.
The Hall-of-Fame Pyramid of players detailed in Bill Simmons’ The Book of Basketball is a great list for us to use as a reference point because it’s well known and the rationale for the rankings was laid out with meticulous detail. So, we can see if our formula is able to recreate the Pyramid and check how our inferred inputs jive with the criteria Simmons’ described in his book.
Simmons’ Pyramid was organized into five tiers: Levels 1 to 4 (ascending order), followed by the top tier, which he dubbed “the Pantheon”. The top three tiers included players ranked 1 to 48:
Simmons discussed each player’s career in a brief profile and each of the bios started with a paragraph summarizing some career stats. Chief among these evaluation metrics was the league-award count — how many MVPs did he win? How many All-NBA teams did he make? Seemingly, Simmons used these awards to sketch out the rough outlines of his all-time ranks. Correspondingly, using our tool, we can max-out the Superiority slider to mimic his thinking.
Those statistical resumes that Simmons used in his book also highlighted the number of quality years which a player turned in (i.e. longevity) and the three-year peak he achieved (i.e. peak form). Any player who didn’t check both of those boxes — because “…they never peaked for two or three years as a top-five guy…or their careers were shortened by injuries and/or rapidly declining skills…” — were disqualified from admission into the top-three tiers. What’s more, Simmons explicitly expressed his view of the relative importance of these two traits: “I value someone who was great for a short period of time over someone who was good for a long period of time,” he said, prioritizing “…two transcendent years from Bill Walton over fourteen non-transcendent years from Walt Bellamy.” To fit these specifications, we need to make sure both the Peak Form and Longevity sliders are engaged, with Peak having the heavier impact.
Of course, the list was subjective. Simmons, himself, was quick to remind us that “Statistics are extremely helpful, they fill in a lot of holes, but that’s it.”
To bolster the info he gleaned from stats, Simmons repeatedly invoked the mystical powers of “The Secret” as a way to further distinguish between these greatest players of all time. Did Player X understand the Secret? Only Bill knows and the answers transcend box scores. But, we can surmise that “knowing the Secret” more-or-less boiled down to having Team Success. If a player had a lot of playoff success and made big contributions to his team’s winning efforts — he will have racked up a lot of postseason Win Shares and, chances are, Simmons would have considered him an initiated member of the Club of the Secret. As such, we’ll crank that Team Success sucker up to 11 and use it as our surrogate for the Secret.
Finally, to rebuild the Pyramid we need to make some adjustments for era. Simmons indicated that he discredited player achievements from the pre-modern era (i.e. prior to the NBA-ABA merger), especially those from the 1940s and 50s. “Send the likes of David West or Hedo Turkoglu to the early fifties in Doc Brown’s time machine and they’d win four straight MVPs” Simmons joked. Likewise, we can infer from the rankings that Simmons has a somewhat skeptical view of the ABA, so we’re ignoring awards and titles from that league.
Keep in mind, the book (and, thus, the list) was published in 2010. Steph Curry and Kevin Durant weren’t ranked. LeBron James wasn’t in the Top-10. It was a different time. So, to make our formula match, we need to pretty much discount any seasons from the 2010s.
Finally, Simmons is an unabashed HOMER. He loves the Boston Celtics and, to get his hometown heroes Bill Russell and Larry Bird into the Top-5, we’re gonna need to use the Stan multiplier. At the same time, to keep Karl Malone and John Stockton out of the Pantheon, we’re gonna need to penalize all of the Utah Jazz men.
So, what did all this tinkering get us? Check out the results below — we’ve done a pretty decent job of recreating the top-3 tiers of the Pyramid.
We keep the desired top-five (albeit, a bit jumbled up); we keep Bill Russell ahead of Wilt Chamberlain (which was the whole point of the book, as I understand it); and we keep 12 of the 13 Pantheon players in their correct tier.
John Havlicek — riding the Celtics bump up a tier by mistake — was the only Pantheon intruder (he was No. 14 in TBOB) and Moses Malone (No. 13 in TBOB) was the counterbalance who lost his position and slipped into Level 4. Based on the formula, David Robinson deserved to be higher than his TBOB rank of 29th; somewhere in Level 4, perhaps. The algorithm had about half of the Level 3 guys pegged in the right tier, but it also suggested a few Level 2 guys who were potentially worthy of a promotion, including: Dirk Nowitzki (No. 39 in TBOB), Patrick Ewing (No. 40), Gary Payton (No. 41), Jason Kidd (No. 43), Clyde Drexler (No. 44), Dolph Schayes (No. 52), and Robert Parish (No. 59).
*****
Nylon Calculus’ Ben Taylor has his own well-regarded, all-time rankings. And Ben’s list is a next-level challenge for our algorithm, because his criteria really don’t match our levers very well at all. I’ll let him explain how his list works.
“This list will not make traditional “arguments” for players…nor do I care about accolades like All-Star teams…I also don’t care how many rings a player won.” He continues, “This list also goes far beyond the box score — indeed, the box score is merely a reference for quantifying tendencies — so if you’re used to citing points per game and Win Shares, this will be a bit different.” Umm, Ben? You’re not exactly providing a glowing endorsement for our tool, here, man.
In fact, when I described my idea for this project to Ben he told me it was giving him some Uncanny-Valley vibes. He was revolted. He had spent seven years reviewing decades of film to make his rankings and, here I was, trying to reduce all his hard work into a single formula. I could understand why it put him off a bit.
Well, sorry. We’re gonna try to recreate Ben’s rankings, anyway. He made a Top-40 list and the final eight players were bestowed a special distinction. Here they are, the Inner Circle:
Let’s start with the one setting I can actually explain: to recreate Ben’s Top-8, we need to tweak the position dials to favor centers. This makes total sense because Ben’s rankings are wholistic — giving equal credence to offensive and defensive contributions. Our formula, in contrast, has very few inputs that are directly influenced by a player’s defensive abilities. So, the best we can do is use a player’s position as a simple surrogate for his defensive prowess. Because centers tend to be the most impactful defensive players, we need to make sure that we are prioritizing centers to recreate his list. So then, that part checks out; but it’s pretty much the last thing that does.
Ben said he doesn’t care about rings; but to recreate his Top-8, we need to max-out the #Rangzzz slider. He said he didn’t care about Win Shares; but to re-form his Inner Circle we need to flip the switch on Team Success, too. He said he didn’t care about Awards; but to remake his list, we need to juice the league awards trait.
Ben said he had no interest in putting players in a time machine — that “It’s about the impact each had in his own time over the course of a career.” But, to have the algorithm spit out the right answers, we need to discount all stats from before 2000.
Ben said his list was objective, but to get KG into the Top-8 we need to impose a Boston Celtics bias — just like we did for Simmons!
Oh — and we also let Kobe slip into the Top-8. Ben is going to be so mad.
*****
For the last comparison, we turn to a list from Ball Don’t Stop.
https://twitter.com/balldontstop/status/1039655803290935296
This list is another great litmus test for our algorithm because it’s totally unique and it has some very distinctive qualifiers. It’s limited to perimeter players and it’s based “purely on ability.” What does this mean? Well, it seems to mean that the guys on this list GET BUCKETS!
To re-engineer this list, all we need to do is max-out the career points-per-game slider and add a few ticks on the Superiority slider. Team success? Efficiency? Championships? None of that matters.
Again, this list was for perimeter players only — so we need to prioritize guards and downplay centers. You’ll probably also notice that this Top-10 is missing a lot of familiar faces from the old days. Jerry West? Larry Bird? Magic Johnson? No thanks. Remember, we’re talking about PURE ABILITY, here. So, we’ll go ahead and zero-out all of the decades before the 1980s and voila: Kobe, Jordan, LeBron, AI, KD in the Top-5, right where they belong. Steph, Harden, D-Wade, and Melo not far behind and all in the Top-12.
OK — so you’ve seen what the tool can do. Now it’s time for you to go out and build your own Top-50 list. Have fun and remember: there are no right or wrong answers, here; as long as you find a list that’s right for you!