Amalgamated Position Index for NBA Players

Al Horford

Although basketball has always been one of the most fluid and dynamic sports, the recent wave of small ball has made assessing player positions with any sense of objectivity even more difficult. The fact of the matter is that, unlike baseball, football and hockey, there’s no set framework that rosters need to be built around. Coaches have historically followed certain patterns but even those have been thrown out the window as big men have been asked to defend out to the perimeter and everybody else on the floor is expected to shoot .400 from beyond the arc.

To be honest, we’ve never been able to consistently point to a position on the court and say “that must be the small forward” but now it’s outright impossible. Do we even use small forwards anymore? Do basketball teams run two guards, two forwards and a center? Or one point guard, two wings and a pair of post players? Heck, can we at least agree to preserve the 1-5 scale as our best bet of labeling players in terms of their mobility relative to each other? It’s exhausting and irritating but your guess is as good as mine. This is the world we live in and somehow, deep down, it’s Don Nelson’s fault.

With that out of the way, there’s an opportunity here to salvage our misappropriated player position framework and at least attempt to come up with a new standard. There are resources like that track five-man combinations from which hoops fans can infer player positions but even those hit a snag when somebody like Dwane Casey trots out Cory Joseph to play in the Raptors backcourt alongside Kyle Lowry. Is Joseph the shooting guard? Are they both still point guards?

With this newfound suspicion that everything we thought we knew is a lie, it’s hard to take the player positions put forth by publications like ESPN and Yahoo at face value. They’re subjective placeholders that reveal only that each player falls into at least one of several previously established player archetypes to which the dude updating profiles at that moment in time subscribes. So what makes Anthony Davis a power forward and DeMarcus Cousins a center? What made Tim Duncan a power forward other than the fact that David Robinson was already firmly entrenched at the five when he showed up to training camp in 1997? Does the guy who takes the jump ball automatically get branded the club’s pivot? The tallest guy? The slowest big man?

Don’t answer any of these questions. The point is that there’s no consensus.

What I’ve decided to do when coming up with some standardized position database is take the amalgam of six different prominent basketball publications, aggregate the various positions each player is listed at and then pump out one result. The idea here is that by balancing all of the various publications and their inherent subjectivity – not to mention their varying formats – we’re hedging our bets against any weird outliers. For instance, get this: ESPN lists Amir Johnson as a power forward. Harmless enough, isn’t it? Yahoo, however, pipes in and says that he’s more of a PF/C. But wait, CBS says the guy plays the three. Crazy, right? Well, not exactly, Basketball-Reference chimes in and says that he plays both the three and the four. Rotoworld? They follow suit with Yahoo and add that he’s a PF/C.

Rather than try to piece together which ideology each publication is basing their decision on, why not dump them all into one pot and then extract some semblance of common ground. Let’s protect ourselves against weird deviations by combining all the different results together. Consider it an exchange-traded fund for player positions, something that generally encompasses the trends of the market without flying off the rails with any particularly noticeable deviation.

I present, the Amalgamated Position Index (API), a weighted summary of six publications and the one resulting player position to rule them all.

The Publications
There are several reasons why I settled on the six publications that I settled on. Of them, two strictly adhere to the traditional 1-5 scale (ESPN and CBS), two use the 1-5 scale but aren’t afraid to hybridize when appropriate (Yahoo and Basketball-Reference) and two use the abbreviated GFC scales (Rotoworld and, also allowing for hybridization.

Yahoo: PG, G, SG, GF, SF, F, PF, FC, C (G interpreted to mean PG/SG, F interpreted to mean SF/PF)
Reference: PG, PG-SG, SG, SG-SF, SF, SF-PF, PF, PF-C, C*
Rotoworld: G, GF, F, FC, C G, GF, F, FC, C

 *Note that B-R actually lists a primary position, meaning there would be a difference between SF/SG and SG/SF. This, however, does not impact our tool. Another thing to note is that B-R doesn’t shy away from triple eligibility either, in these cases we consider their primary position and secondary position so long as the two are adjacent on the 1-5 scale.

The Methodology
From each publication I infer a player’s position on the 1-5 scale. For four of the sites, this is simply a matter of converting PG into 1, SG into 2, etc. If a player is listed at two positions, a PG/SG for example, I input 12. (Don’t think of these as numbers so much as characters in a string, 12 represents 1 and 2).

Where this gets complicated is with Rotoworld and While it’s easy enough to convert C to 5, things get complicated there. If one of these publications lists a player as a GF, spanning both the guard and forward positions, I make the assumption that they don’t mean Jimmy Butler plays point guard and power forward, rather that he bounces between shooting guard and small forward. That’s an easy 23. Similarly, when a player is listed as an FC, it’s inevitable that the player is a PF/C and not an SF/C.

The prickly part, however, is when a player is listed only as a G or an F, with no other clue to decipher what exact position on the 1-5 scale they refer to. A G could be a point guard or a shooting guard and I wouldn’t want to conflate the two and give them equal weighting. We wouldn’t want to bastardize our results by adding a value that could suggest Chris Paul was a shooting guard. As a result, in cases where a player is listed as a G or an F, that position is taken at half value in the final calculation.

Case Study: Al Horford

Once I have uniform values for each of the six publications, I can start comparing and contrasting them all. Let’s use Al Horford as an example, a player who at first glance seems to bridge the PF/C positions.

With our first two results, we’ll see what publications are saying when they’re not allowed to hedge their bets and pick multiple positions.

ESPN: C -> 5
CBS: C -> 5

Hmm, from this we can see that neither are influenced by the fact that Horford shared a frontcourt with Zaza Pachulia during his tenure in Atlanta. In that scenario, Horford was almost indisputably the PF. Already we’re treading into uncomfortable territory. Validating Horford as a pure 5 is the fact that 82Games suggests as much. Unfortunately, ESPN‘s own depth charts contradict this. What do we learn from this? Nothing new. Player positions are a gong show subject to interpretation.

Let’s continue by bringing Yahoo and Basketball-Reference into the fold, two publications that aren’t afraid to give themselves the editorial leeway of submitting multiple positions when necessary.

Yahoo: FC -> 45
B-R: C -> 5

Here we see that Yahoo is the first to cave and say that he plays both. Basketball-Reference, however, holds firm that Big Al only plays center these days, even though they, as a publication, happily assign PF/C tags where they deem appropriate. At this point, the majority lists Horford as center, not a power forward nor a PF/C hybrid.

Let’s see if Rotoworld or can tilt the scales.

Roto: FC -> 45
NBA: C -> 5

This does it. Our grand total reads 5, 5, 45, 5, 45, 5. Six out of six publications agree that Horford sees time at the five. Two out of six publications agree that he sees time at the four. A six to two split seems pretty indicative but let’s look a bit closer. If you account only for publications that allow for hybridization, half of them (two of four) do so. There’s a case to be made that Horford is a PF/C hybrid, we’re just not making it… yet.

The threshold that I’ve decided to implement is a simple one, if three of six publications say a player has eligibility somewhere, I’m not going to disagree with them. This is the cut off for what I’ll use as my Standard API (Amalgamated Position Index).  Going by this, Al Horford is a C and only a C. Sorry Al.

I’d like to allow for some breathing room however, especially if fantasy basketball sites like want to use the methodology for determining eligibility. Therefore, I’ve added in secondary player attributes. In addition to a player’s Standard API, I’ve added Liberal APIs (wherein just two of the six publications have to confirm a position) and Conservative APIs (wherein four of the six publications have to confirm a position).

Suddenly those two PF/C nods are enough for Horford to qualify as a PF/C.

Case Study: Jimmy Butler
To drive home some different mechanisms at play here, let’s take a look at Jimmy Butler. We’ll speed things along a little, too.

ESPN: SF -> 3
CBS: SF -> 3
Yahoo: GF -> 23
B-R: SG/SF -> 23
Roto: GF -> 23
NBA: F -> 34(!)

You’ll see a few things at play here, let’s start at the bottom with our first example of the weighted adjustments to Rotoworld and results. In cases such as this, where lists Butler as a forward, we note that the general F position puts out here would cover both SF and PF. I don’t feel comfortable with that knowing that this is likely more a casualty of the simplified GFC format than a genuine product of the traditional 1-5 scale. I don’t think anybody at would make the claim that Butler shares time at the three and the four.

This is where the concept that I discussed above comes into play. When Rotoworld or list a player as a G or an F, we have several options. Technically a G label would indicated PG/SG and an F label would suggest SF/PG. This is troublesome. What can we do though? Well, we can try to predict if they mean one and not both of the positions. I’d rather not do this, considering that we’re going out of our way to take subjectivity out of this process as much as possible, let’s keep it out of it. Alternatively we could count this as a credit to both positions, an F, then would be akin to Basketball-Reference listing the player as an SF and a PF. I disagree with this, especially considering that a player like Chris Paul is listed as a G by and Rotoworld. Should these count as two legitimate SG nominations? Of course not, that right there would be enough give him a Liberal API of PG/SG.

What I ended up doing is halving the credit a player gets when listed by or Rotoworld at either G or F. You’ll notice that the final tally for Butler, then, reads as follows: 3, 3, 23, 23, 23, 34. Here Butler is listed as a 2 just twice. He’s listed as a 3 in all six scenarios. He’s listed as a 4 in one scenario all because of that pesky F value. When we halve the value, however, we see that he’s listed as a 3 in 5.5 scenarios. A 2 in three scenarios and, yes, a 4 in 0.5 scenarios.

This complexity isn’t ideal but I’d rather live with the complexity rather than get rid of the GFC format altogether, and Rotoworld often end up breaking ties and are a valuable way of accommodating for the fact that ESPN and CBS refuse to acknowledge dual eligibility even in the most obvious of cases.

Another thing we see here is that Jimmy Butler’s final Standard API would be SG/SF (because 5.5 of six publications say he’s a 3 and three of those publications say he’s also a 2). His Liberal API, unsurprisingly, would also be SG/SF.

If you were looking for the most strict interpretation of his eligibility, however, you would see that his Conservative API is just SF. Why? Well, could have easily listed Butler as a GF but they didn’t. That’s telling right there and indicative of the value and Rotoworld bring to the process despite their use of the abbreviated GFC format. By taking a more conservative approach here, we can also see that neither ESPN nor CBS considered making him a SG primarily, opting instead to classify him as an SF. The Conservative API is basically a way of stripping away any and all hybridization when there’s even a shadow of a doubt that a player should qualify for multiple positions.

Ultimately, when it comes to player positions, there’s a shroud of ambiguity that makes it difficult for any one individual or publication to pinpoint a player. What else can we realistically expect? We’re trying to put a label on something intangible and the job description for a modern day small forward will change depending on whomever you ask. Given the inherent subjectivity with this exercise, the only safe way to encompass all the varying views is to include the varying views in one standardized account.

Perhaps, in this day and age, positions are largely irrelevant. That’s a notion that certain NBA coaches would have you believe. We still, however, need a way of classifying players if we plan to discuss them and categorize them based on shared attributes. Knowing this, this proposed Amalgamated Position Index is a comprehensive and entirely objective way of consolidating all of the existing opinions published by prominent basketball publications into one workable result.

About the author

Austin Kent

Austin Kent is the Editor-in-Chief of The Good Point and the Network.