The NBA Draft and the Limits of Prediction
Sophia Rodriguez
2026-03-16 ·
Every June, the NBA draft produces a strange mixture of confidence and speculation. Teams speak about prospects with the language of certainty—“safe pick,” “high floor,” “franchise cornerstone”—as if the future of an eighteen- or nineteen-year-old player were something that could be measured with enough film, data, and scouting reports.
Yet draft history quietly resists that tone. A skinny teenager from a small Greek league becomes a two-time MVP. A second-round center selected during a Taco Bell commercial turns into the best passing big man the sport has ever seen. Meanwhile, a dominant college center chosen first overall sees his career dissolve under injuries that were impossible to fully forecast.
The draft looks analytical on the surface, but its deepest judgments belong to a different category of thinking. The economist Frank Knight once drew a famous distinction between riskRisk, in Knight’s framework, refers to situations where outcomes are uncertain but their probabilities can be estimated from past data—like an insurance company calculating accident rates. and uncertaintyUncertainty, as Knight defined it, describes situations where the possible outcomes themselves are not fully known, or where the forces shaping the future are too complex to assign meaningful probabilities. , and the difference between those two ideas helps explain why the NBA draft remains one of the league’s most fascinating intellectual exercises.
Risk and the Unknown Future of a Player
Knight argued that some futures can be estimated through probability while others cannot. Risk refers to situations where outcomes are uncertain but still measurable. Insurance companies, for example, can estimate the likelihood of accidents because enough past data exists to produce stable patterns.
Uncertainty is something more difficult. It describes situations where the relevant outcomes themselves are not fully known, or where the causal forces shaping the future are too complex to model reliably. In those cases, probability calculations give the appearance of precision without fully capturing the reality of the decision.
Draft evaluation sits directly on that border.
Some parts of a prospect profile resemble risk. Height, wingspan, age, shooting percentages, turnover rates, and defensive activity can all be studied across large datasets. Teams can reasonably estimate how often players with certain statistical profiles succeed.
But the most important questions about a young player belong to the second category. Will he add a reliable jumper? Can his body handle the physical demands of the NBA for a decade? Will his feel for the game expand once he is surrounded by elite teammates? How will fame, money, and pressure reshape his work habits and identity as a player?
Those are not simply difficult questions. They are questions whose answers do not yet exist.
Drafting, in other words, is not merely about evaluating what a player is. It is about judging what kind of future he might grow into.
The Giannis Problem
Consider Giannis Antetokounmpo in the 2013 draft.
At the time, he was a long, intriguing teenager playing in the Greek second division for Filathlitikos. Scouts saw flashes of ball-handling, defensive instincts, and unusual fluidity for his size. But the picture was incomplete. His body was still developing, his competition level was unfamiliar to many NBA evaluators, and his offensive role had not yet taken a clear form.
Milwaukee selected him fifteenth overall, and even that pick was considered adventurous.
What makes the Giannis story interesting is not simply that teams underestimated him. The deeper point is that his eventual form as a player was not yet clearly describable. He grew several inches after entering the league, developed into a full-court attacking force, and eventually became both a two-time MVP and a Finals MVP.
Calling him “raw” misses something important. Raw prospects still resemble known developmental templates. Giannis represented a future that the existing templates did not quite capture.
Knight would describe that situation as uncertainty rather than risk. Evaluators were not choosing between well-understood probabilities. They were trying to imagine a basketball player who did not yet fully exist.
The Case of Nikola Jokić
The same pattern appears in the story of Nikola Jokić.
When Denver selected him forty-first in the 2014 draft, he was already a highly skilled offensive player in Europe. He could pass, score with touch around the rim, and read the game at a level unusual for a center. What he lacked was the athletic profile that scouts typically associate with future stars.
The difficulty was not a lack of evidence. Teams had film and statistics. The difficulty was interpretive. The existing mental model for NBA greatness at the center position emphasized physical dominance, rim protection, and vertical explosiveness.
Jokić fit none of those expectations.
His rise into the lowest-drafted MVP in league history revealed something about the limits of draft models. The information was available, but the framework used to interpret that information undervalued the possibility that a center could control an offense primarily through decision-making and passing vision.
This is where another philosophical insight quietly supports Knight’s idea. David Hume once pointed out that our expectations about the future rely heavily on inductionInduction is the reasoning process of drawing general conclusions from specific past observations. Hume famously argued that inductive reasoning rests on habit rather than logical certainty—past patterns do not guarantee future ones. from the past. If we have seen a certain type of player succeed before, we assume similar players will succeed again.
But resemblance is not certainty. It is only evidence.
The draft constantly confronts that problem. Teams search for players who resemble successful archetypes, even though the next great player might arrive in a form the archetype does not yet describe.
Consensus and the Limits of Safety
The famous 2007 draft decision between Greg Oden and Kevin Durant illustrates a different side of the same philosophical problem.
Oden was widely viewed as the safer choice. He projected as an elite defensive center with dominant physical tools, and historically players of that type had anchored championship teams. Durant, by contrast, raised questions about his slender frame and defensive projection.
With the benefit of hindsight, the choice looks disastrous. Durant eventually accumulated more than 180 career win shares while Oden’s injury-ravaged career produced only a small fraction of that value.
But hindsight disguises the structure of the decision itself.
Even at the time, teams could recognize medical uncertainty around Oden. What they could not do was assign reliable probabilities to how that uncertainty would unfold over a decade. Long-term health is one of the clearest examples of Knight’s distinction. Scouts may identify a concern, but the range of possible outcomes remains fundamentally unpredictable.
Durant did not simply turn out better than expected. The future of both players contained elements that evaluation methods could not fully measure.
When Development Changes the Story
A more recent example emerged in the 2017 draft, where Markelle Fultz went first overall while Boston traded down and selected Jayson Tatum third.
Before the draft, Fultz appeared to be a complete offensive guard. His scoring ability, playmaking, and shooting profile suggested a natural translation to NBA offense. Tatum was also highly regarded, but the consensus placed Fultz at the top of the board.
Then the uncertainty of development intervened.
Fultz’s early NBA career was disrupted by injuries, shooting struggles, and a loss of confidence that reshaped his role and trajectory. Tatum, meanwhile, immediately stabilized Boston’s offense as a rookie, playing eighty games and shooting over forty-three percent from three-point range.
The difference between those careers cannot be explained simply by saying scouts “missed” a skill. Instead, the interaction between health, mechanics, psychology, and organizational environment altered the path of each player in ways that were nearly impossible to foresee on draft night.
That is precisely what uncertainty looks like in practice: a future shaped by variables that reveal themselves only after the decision has already been made.
Betting on the Unknown
The 2023 selection of Victor Wembanyama offers a final perspective on the same philosophical landscape.
Wembanyama entered the league as perhaps the most celebrated prospect in years, a 7-foot-4 player with guard-like skills and defensive reach that seemed almost fictional. The San Antonio Spurs selected him first overall with little hesitation.
Yet even that apparent certainty existed alongside enormous unknowns. Players of that size rarely combine durability with long careers, and the league had never seen quite this combination of height, mobility, and perimeter skill.
What the draft revealed in that moment was not certainty but judgment. Teams were willing to embrace a significant degree of uncertainty because the potential reward was historically large.
The decision was not that Wembanyama’s future could be predicted with precision. It was that the upside justified the risk of the unknown.
Seeing the Draft More Clearly
Once the distinction between risk and uncertainty becomes visible, the NBA draft begins to look slightly different.
Mock drafts and prospect models often present evaluation as a ranking problem: which player is better right now, and which profile historically succeeds more often. Those tools are useful, but they operate most effectively in the parts of the process that resemble measurable risk.
The true challenge of drafting lies elsewhere. Teams must judge how a young player’s body, psychology, skill set, and environment will evolve over the next decade, even though many of those forces are still forming.
That is why the language of the draft often drifts toward narrative. A “safe pick” is usually the player whose future role is easiest to imagine, not the player whose outcome is genuinely most predictable. Consensus rankings frequently reflect shared comfort rather than superior knowledge.
The best front offices understand that difference. They try to separate what can be estimated from what must simply be interpreted, and they remain aware that some parts of the decision will remain unknown until years later.
In that sense, drafting a player resembles a broader philosophical exercise. It requires epistemic humilityEpistemic humility is the recognition that one’s knowledge is limited and that confident judgments may rest on incomplete evidence—a stance that encourages openness to surprise and revision. about the limits of prediction, patience with incomplete information, and the willingness to act despite both.
The draft board may look like a list of prospects, but it is really something else: a map of the futures that organizations believe they can see, and the uncertainties they are willing to accept.
Footnotes / Philosophy Terms
1. Risk ↩
Risk, in Knight’s framework, refers to situations where outcomes are uncertain but their probabilities can be estimated from past data—like an insurance company calculating accident rates.
2. Uncertainty ↩
Uncertainty, as Knight defined it, describes situations where the possible outcomes themselves are not fully known, or where the forces shaping the future are too complex to assign meaningful probabilities.
3. Induction ↩
Induction is the reasoning process of drawing general conclusions from specific past observations. Hume famously argued that inductive reasoning rests on habit rather than logical certainty—past patterns do not guarantee future ones.
4. Epistemic humility ↩
Epistemic humility is the recognition that one’s knowledge is limited and that confident judgments may rest on incomplete evidence—a stance that encourages openness to surprise and revision.