Possession Maximizers

Photo credit

Something that’s been on my mind recently is the tension between short-term and long-term thinking, and how often individually rational short-term decisions produce collectively worse outcomes over time. You can see this everywhere. One example is the way we build cities.

I live in Columbus, the state capital of Ohio. Columbus sits near the center of the state and geographically speaking, has very few natural constraints on its growth. Compare that with Cleveland, where Lake Erie creates a hard boundary to the north. Or Cincinnati, where the Ohio River constrains the city to the south and steep terrain creates additional barriers to development. Those constraints matter.

Historically, waterways were enormous economic advantages, which helped cities like Cleveland and Cincinnati develop earlier and more densely. But the constraints that came with those waterways also limited where those cities could expand. When outward expansion becomes more difficult, development is forced to make better use of the land that already exists.

Columbus has never faced the same pressure. There is almost always another piece of cheap land farther from the city center. That difference creates an interesting incentive problem.

Imagine you are a developer trying to build 100 apartments. In Columbus, one option is to acquire expensive land in or near downtown, navigate more strict permitting, demolish or redevelop whatever is already there, and construct a relatively expensive multi-story building.

The other option is to simply drive farther out from downtown, buy a cheaper and largely empty parcel of land, and build there.

In the short term, the second option is usually easier. The land is cheaper. Construction is simpler. There may be fewer neighboring property owners to deal with. You often don't have to demolish anything first and because Columbus can expand in every direction, there is almost always another parcel available.

Do that once and it doesn't matter very much. Do it thousands of times over several decades and you get suburban sprawl.

Columbus now occupies an enormous geographic footprint, and its functional metropolitan area extends far beyond the city limits. A large share of the people who participate in Columbus economically and socially don't actually live within the City of Columbus at all. They live in surrounding municipalities and townships connected to the city by highways, employment, shopping, and commuting patterns. Columbus isn't unique in this respect. You can see similar development patterns in many American cities where geography places relatively few limits on outward growth. But the important question isn't simply why does sprawl happen?

It's why perfectly rational people keep making decisions that, when repeated across an entire society, can create worse long-term outcomes.

The developer isn't necessarily doing anything wrong, depending on your perspective of course. If you have a limited amount of capital and you need to build 100 apartments, why would you voluntarily choose the project that costs more, takes longer, involves more regulatory uncertainty, and carries a greater chance of failure?

The downtown high-rise might create more long-term value. It might use infrastructure more efficiently, place residents closer to jobs and amenities, and contribute more to the density and vitality of the city. But those benefits are spread across decades and across thousands of people. The developer absorbs the risk today. That's the problem.

The incentives favor the short-term decision even when the long-term outcome may be worse¹.

And the amount of risk you can tolerate depends heavily on how much capital you already have. If you have $1 million available to invest, your decision-making is going to look very different from someone with $100 million. The person with $100 million can afford to wait longer, absorb setbacks, and pursue projects whose rewards may not materialize for years.

The person with $1 million often can't.

That difference between what is optimal in the short term and what is optimal in the long term is what I want to explore.

The more I thought about this, though, the more I realized the interesting variable wasn't really time. It was constraint. When easy alternatives are abundant, there's less incentive to find a harder solution. Constraints remove those alternatives and force optimization. And that distinction has an interesting parallel in how basketball players develop.

When it comes to evaluating basketball players, the collective basketball community has gotten increasingly precise. We split hairs now in ways we probably couldn't have 20 or 30 years ago. We have more film, more data, more exposure to players at younger ages, and more language for describing what they're actually good at. But I think there is another step to take.

A concept related to short-term versus long-term thinking, at least in my head, is what I've been calling a polyspectrum view of player evaluation. I don't know if I love that word yet, but the idea is pretty simple.

At the most basic level, you can evaluate a player as good or bad. That's binary. Then you can put everyone on a spectrum: How good is this player relative to everyone else?

But even that collapses too much information into one dimension.

Two players can produce roughly the same amount of value while arriving there through completely different mechanisms. They can have different physical tools, different cognitive advantages, different skill sets, different weaknesses, and radically different dependencies on all of those things. So instead of one spectrum, I think we should be thinking across several of them simultaneously.

The example I keep coming back to is the difference between the brilliance of LeBron James and the brilliance of Chris Paul or James Harden.

All three are basketball geniuses. All three are Hall of Fame players. But their brilliance manifests itself very differently.

Paul and Harden are, to me, possession maximizers.

A huge part of their greatness comes from their ability to squeeze as much value as possible out of individual possessions. Harden's version of this is most visibly associated with the step-back three that eventually spread across the league, but it goes much deeper than that. It's the fouls. The pacing. The manipulation of help defenders. The refusal to waste movement. The ability to repeatedly engineer the same advantageous decision until the defense proves it can stop it.

This was a breakdown I helped Kirk Goldsberry write and research when I was at ESPN. Harden’s step-back and the heliocentric offense concept is what I would attribute as a precursor to my possession maximizer concept.

Paul's version looks different, in part because he's barely six feet tall. He became one of the greatest point guards ever while operating with a physical limitation that most historically great NBA players simply didn't have. He became absurdly good at manipulating defenders, creating fouls, controlling pace, protecting the ball, and making shots from areas of the floor that modern basketball would generally prefer you avoid.

Their constraints helped create their games, and this is where the city-planning analogy comes back:

Chris Paul and James Harden are Cleveland and Cincinnati. LeBron is Columbus.

Paul and Harden had harder boundaries around the ways they could dominate an NBA game. Those boundaries forced them to make extraordinarily efficient use of the tools available to them.

LeBron had more land.

By the time LeBron was 16 or 17 years old, there was almost no realistic question that he possessed physical advantages unavailable to nearly everyone else who had ever played basketball. His combination of size, speed, strength and coordination gave him a much larger solution space.

That doesn't mean he didn't work as hard, or that his intelligence mattered less. Quite the opposite: part of what makes LeBron perhaps the most complete basketball player ever is that the basketball genius sits on top of that physical foundation, but he wasn't forced to optimize every basketball problem in the same way Paul was.

Paul could not decide one night that the solution was simply to be 6-foot-9, 260 pounds and run through the defense. LeBron can.

That distinction matters when evaluating players.

We already understand some version of this with prospects. Nobody watching teenage LeBron thought, "He's a highly rated recruit, just like these other highly rated recruits.”⁷ The magnitude of the difference was obvious. College coaches weren't seriously recruiting him because everyone understood that college basketball was almost certainly irrelevant to his trajectory.

Paul and Harden were also elite high-school players, but they were still normal elite recruits in the sense that college basketball was a necessary and an important part of their development. For some reason though, once players become established NBA stars, we tend to compress those differences.

We start talking about "superstars" as though they all occupy roughly the same place. We debate whether the sixth-best player in the league is better than the eighth-best player without always asking the more interesting questions: Why is this player this good? What combination of advantages is producing the value?

On one end, you can have someone like Chris Paul, whose physical limitations require an almost ridiculous level of skill, processing and possession-by-possession optimization to reach an all-time level.

On the other end, you can have someone like Steven Adams or Mitchell Robinson, whose exceptional physical traits allow them to create substantial NBA value without needing anything close to Paul's offensive skill set.

Neither description is an insult. That's the point. They're different planes.

And somewhere far out in several directions simultaneously you get someone like LeBron, whose greatness has never depended on a single overwhelming advantage because there are so many of them stacked together.

That's why I'm not sure evaluating players on one spectrum is enough: How good are they? I'd rather ask:

What makes them good?

How scarce is that advantage?

How dependent are they on it?

What happens when an opponent takes it away?

How much does the advantage scale from one possession, to one game, to one playoff series, to an entire career?

And how many alternative ways does the player have to create value when his preferred method stops working?

That last question is where the short-term versus long-term distinction becomes especially interesting.

Possession maximization is enormously valuable. If you can repeatedly find the highest-value action available right now, you can build an incredible basketball career doing it.

But maximizing every individual possession is not necessarily the same thing as maximizing your season, or your career. That doesn't mean Chris Paul or James Harden somehow failed.

In fact, I think viewing their careers primarily through what they didn't accomplish dramatically undersells what they did accomplish.

Chris Paul is barely six feet tall and has a legitimate place in discussions involving the greatest point guards ever². Even before getting into championships or playoff exits or any of the other things we use to judge careers, that fact by itself is absurd.

His career is almost a case study in how far intelligence, skill and optimization can stretch the boundaries imposed by physical constraints.

Harden is slightly different, but similarly interesting. He entered the NBA in an unusually crowded environment: Oklahoma City had already drafted Kevin Durant and Russell Westbrook before selecting Harden third in 2009. Harden became the Sixth Man of the Year in his third season, helped Oklahoma City reach the 2012 NBA Finals, and was then traded to Houston before the following season.

So instead of spending too much time asking whether Harden's career ultimately produced the team outcome we expected, I'm more interested in what his career tells us about player development.

He wasn't born with LeBron's menu of physical solutions. So he built an extraordinarily specialized set of basketball solutions instead³.

And that may be the more useful way to evaluate the next generation of players. Don't just ask how good a prospect is. Figure out why they’re good, what constraints they’re solving for, and how many different ways they will eventually be able to solve them.

If you come at sports from a statistics background, you've probably thought about some version of this before. What is the maximum amount of value you can extract from a single possession?

You want a three. You want to get fouled while taking it. You want to score as quickly as possible so you can create more possessions. On defense, you want to force a turnover immediately and get the ball back before your opponent can even attempt a shot. If the objective is simply to maximize points over a finite period of time, the theoretical strategy becomes pretty obvious: create as many possessions as possible and maximize the expected value of every one of them.

You can see exaggerated versions of this logic in video games. In Madden, for example, certain mathematical approaches might tell you to go for two far more often than conventional football wisdom would suggest. If you're playing entirely to maximize expected points, why voluntarily leave value on the table? The problem is that basketball isn't played in a spreadsheet. How does this practice work in reality? My last article pays heed to that.

You could decide that getting fouled is incredibly valuable and therefore try to draw a foul on every possession, but you're not actually going to get the call every time. The harder you optimize exclusively around drawing that call, the more predictable you become. Defenders adjust. Officials adjust. Playoff opponents spend seven games studying the exact thing you're trying to exploit.

At some point, maximizing an individual edge starts creating dependencies on that edge, and a practical version of extreme possession-by-possession optimization starts looking a lot like Chris Paul or James Harden.

That's not a criticism of either player. It's almost the opposite. They became historically great by identifying advantages available to them and extracting an absurd amount of value from those advantages, but there is a difference between maximizing the value of the solution you have and having a huge number of solutions available in the first place.

It obviously looks more like LeBron James. Michael Jordan belongs in this conversation too, along with the very small number of players whose games were so complete that opposing teams rarely had one structural weakness they could simply attack until it broke.

LeBron is the clearest example for me because his advantage isn't that he's literally the best player ever at every individual basketball skill. He doesn't need to be.

Maybe he's in the 99.999th percentile when you combine physical ability, skill, intelligence, processing, durability and adaptability, but the important part isn't that he's 100th percentile in every individual category. It's that he's extraordinarily high in almost all of them at the same time.

And as one advantage has diminished with age, he's been able to redistribute his game toward the others. That’s what I mean by optionality. LeBron doesn’t need the highest-value answer to remain the same because he has enough alternative answers to survive when circumstances change.

It isn't necessarily maximizing the expected value of the possession directly in front of you. It's having enough resources that you don't have to, and can afford to have a long term outlook.

This brings us back to the real-estate example. LeBron is the developer with $100 million. Chris Paul is the developer with $1 million.

The developer with $1 million has to be incredibly disciplined. He has to find the favorable parcel, minimize construction costs, avoid unnecessary risk and extract as much value as possible from every dollar. One mistake might destroy the project. The developer with $100 million can do all of those things too, but he doesn't have to.

He can buy the expensive land. He can wait. He can absorb a mistake. He can survive a project going over budget. He can pursue an opportunity whose payoff won't arrive for ten years.

Most importantly, he has optionality. And I think optionality is one of the things we're still not very good at accounting for when we evaluate basketball players.

After writing all of this out, I realize there's a fine line between the point I'm trying to make and simply saying, "Yeah, no shit, LeBron is better than Chris Paul and James Harden." That's not really the point.

The point is that Chris Paul and James Harden were never particularly likely to have careers that looked like LeBron's, and that should have been apparent well before we knew how any of their careers would end.

With Paul, it was obvious. He's barely six feet tall. With Harden, it was less obvious, but the important distinction isn't that Paul and Harden have narrow skill sets. They don't. They're two of the most skilled offensive players ever.

What's narrower is the range of underlying advantages through which they can dominate a basketball game, and that matters for almost everything that comes afterward. It matters for how they age. It matters for how opponents can attack them in a playoff series⁸. It matters for how much a rule change or officiating change affects them. And maybe most importantly, it matters for how you have to construct a roster around them.

We talk constantly about "building around" stars, but not every star imposes the same constraints. The fewer ways your star can create value, the more specific the pieces around him need to be.

If your star dominates with the ball but provides relatively little value without it, you need players who complement that. If he can be attacked defensively, you have to compensate for that. If he can't shoot, you need shooting somewhere else. If he's small, you need size elsewhere.

Every important constraint on the star creates a problem the roster has to solve somewhere else. That's where someone like Victor Wembanyama gets really interesting.

Being too tall is certainly preferable to being too short in basketball, but Wembanyama's physical profile is so extreme that the Spurs still have to think about nearly every roster decision through the question of how that player interacts with him. They were already doing that because of how great he already appears to be, but don’t confuse that trajectory with the trajectory of LeBron’s career. That isn't necessarily a bad thing. Building around Victor Wembanyama is a tremendous problem to have.

But there is a distinction between acquiring the most talent possible and figuring out how it fits later and acquiring the best talent possible within an ecosystem built around one specific player.

Both can result in championships. But the second is more constrained than the first. Think about the 2010-11 Heat. LeBron James and Dwyane Wade had obvious redundancies. Neither was an especially threatening three-point shooter at that point. There were possessions where you could almost see the geometry getting awkward.

Those problems were real, but Miami could tolerate them because both players brought so many other things to the table. Their overlapping weaknesses constrained the team, but their breadth gave Miami a huge number of ways to solve those constraints.

That's optionality, you know the thing we talked about Brad Stevens striving for in my previous article.

And it's what makes Wembanyama such a fascinating case study for this entire idea. For a long time, I thought about Giannis Antetokounmpo similarly.

On several of the big career milestones we tend to use when talking about all-time trajectories, Giannis was keeping pace with LeBron surprisingly deep into his career. He won an MVP at 24. He won a championship and Finals MVP at 26, and none of that seemed remotely inevitable when he was drafted.

There have been plenty of prospects since then where some version of that logic has been applied and nothing remotely Giannis-like happened. Aleksej Pokuševski is an obvious example.

Giannis happened to grow, add an enormous amount of functional strength and retain enough coordination and mobility that his physical development expanded his game rather than destroying it. He became something close to a thought experiment:

What if you gave a Shaq-like interior force the freedom to grab the rebound, bring the ball up himself and initiate the offense?

For years, I kept waiting for something to derail it.

I thought the physical transformation might eventually push him toward an Anthony Davis-like career arc, where availability becomes an increasingly important part of the conversation. I wondered whether the combination of size, explosiveness and workload could really remain sustainable. For the most part, I was wrong.

And now I find myself asking some of the same questions about Victor. You're telling me a guy that tall, who moves like that, isn't eventually going to spend significant portions of his career hurt?⁴

Maybe not. Maybe his coordination is itself such an outlier that the normal assumptions don't apply. That uncertainty is part of what makes him so interesting.

Victor simply isn’t just a 100th-percentile version of the possession maximizer, though.

I think he's potentially something much stranger. He may already exist near the extreme end of multiple spectrums simultaneously. His height is an outlier. His length is an outlier. His coordination relative to his size is definitely an outlier. His skill relative to his size is an outlier.

So the question isn't merely how efficiently he can exploit one enormous advantage.

The question is how many additional solutions he can develop around it.

Can he play 1,000 NBA games? Can his offensive game keep evolving as defenses learn what they can and can't take away? Can he create elite offense rather than merely finish it? Can he function in several different kinds of lineups? Can his game survive changes in teammates, coaches, schemes and eventually his own body? Those are all essentially different spectrums.

And the answer to those questions will determine whether Victor's career is merely historically great, which already feels like a reasonable possibility, or whether it becomes something we don't really have a precedent for. That's the part of this framework I think becomes useful for evaluating prospects.

Draft evaluation is full of shorthand.

"High floor."

"Low ceiling."

"Raw."

"NBA-ready."

"Upside."

"Old."

They're useful phrases, but they collapse a lot of information. When you spend a month in a draft room dissecting the same prospects over and over, you start to see how easy it is to fall into these categories.

A player who already knows exactly how to create value gets labeled "high floor, low ceiling."

A player with extraordinary physical traits but no idea how to use them gets labeled "high upside."

But I think those labels sometimes confuse how many tools a player currently has with how many different directions his game can still develop.

Chris Paul would have been a ridiculous player to describe as low-ceiling simply because he was small and already knew how to play. James Harden would have been equally ridiculous.

Both were top-five picks. Neither was some safe veteran prospect you selected because you needed a backup guard. They were elite prospects whose existing solutions were already incredibly advanced. So rather than asking only about floor and ceiling, I'd rather ask:

On what planes is this player already optimizing?

What advantages does he currently depend on?

And how many other directions can his game plausibly expand?

That brings me to this past draft.

Two players I viewed as relatively short-term optimizers were Darryn Peterson and Keaton Wagler. Both were fantastic college players, and I think the brightness of what they did well sometimes hid the shadows elsewhere in their profiles.

With Peterson, I think the most obvious path to stardom is that he becomes an absurd isolation scorer. That's different from saying he can't eventually become a true lead guard who organizes an offense, creates easy opportunities for everyone else and consistently adds value defensively. He certainly can, but those are additional spectrums he still has to climb.

The same basic idea applies to Wagler, although the mechanism is different. His immediate NBA value looks more tied to shooting, pick-and-roll creation and making good offensive decisions. Again, that's not a criticism.

The question is whether those existing advantages become the foundation of a broader game or whether they remain the primary reasons they’re valuable.

I initially wanted to put Cameron Boozer in this category too. After writing this entire thing, I don't think I can anymore. Boozer might actually be one of the better examples of why we need this framework in the first place. I think his lack of vertical athleticism clouded my evaluation of him. You look at a 6-foot-8 forward who doesn't explode vertically and instinctively want him to "pop" more. There are certain physical expectations attached to that archetype, and Boozer doesn't satisfy all of them, but he does almost everything else really well.

At Duke, he became the focal point of the offense because he was clearly their best option. He posted players up constantly, probably because it was the easiest way available to him to break down the defense. The off-the-dribble game wasn't what you'd ideally want, and his foot speed obviously contributes to that.

My mistake may have been treating that as evidence of something he can't do rather than something he hasn't needed to do yet. That's an important distinction. If posting someone up works every time, why would you voluntarily stop doing it so you can practice a more difficult solution? The absence of a skill isn't always evidence of an inability to develop it. Sometimes it just means the player hasn't encountered a problem that requires it yet.

Soon, NBA defenses will take some of Boozer's easiest solutions away. Then we'll learn something much more interesting. Does he simply become less effective? Or does being punished create the incentive for another part of his game to develop?

Steph Curry became dramatically better around the rim over the course of his career. Kevin Durant developed a much tighter handle and a devastating crossover. Neither player arrived in the NBA as the final version of himself⁵.

I'm not predicting that Cameron Boozer will follow the same breakthrough developmental path. I'm certainly not holding my breath waiting for him to suddenly become an elite off-the-dribble creator, but I think I was too quick to treat the missing skills as fixed limitations. Maybe the better question is whether the rest of his profile gives him enough optionality to solve problems once those problems actually matter. And that's ultimately what I'm trying to get at with all of this.

Player evaluation shouldn't only be about asking how good somebody is today or projecting how good they'll eventually become on one giant spectrum.

We should be asking why they're good, how many different ways they can be good, how dependent they are on each one, and what happens when the game forces them to find another way. The best prospects aren't necessarily the ones who already maximize every possession. They may be the ones who will have the most solutions available when maximizing the current one stops working.

If you’ve made it this far, please subscribe to this newsletter to get more basketball content like this!