I’m a strong believer that big boards/prospect rankings without any accompanying discussion of player-specific analysis or overall evaluation criteria serve little-to-no purpose in the grand scheme of the discourse. Embarrassingly, I’ve also never posted a big board with either of the aforementioned things. That changes today – starting with a rundown of my philosophy for player evaluation, which centers around targeting (i) unique and (ii) high upside profiles that (iii) reduce the impacts of variance on basketball games. If you want to skip over that discussion, my final 2026 big board will immediately follow, with a rant about the purpose of this whole exercise to cap it all off. It’s been a fun cycle, I hope you enjoy.

Chasing Swans v. Unique Production

Black Swans

Nassim Nicholas Taleb coined the term “black swan event” in his books, Fooled by Randomness and The Black Swan. A black swan event describes “highly improbable events with three characteristics”: 

1. Nothing that has happened before can convincingly point to the possibility of the event
2. High impact
3. Post hoc rationalization to make the event appear more predictable

For the purposes of this post, and as it relates to basketball analysis, black swan event has a much looser definition, one focused primarily on the third and final prong as it relates to over-indexing on unpredictable, unrepeatable prior events in one’s future decision-making. 

In real-life, black swan events like 9/11 and Pearl Harbor created reactive responses (e.g. PATRIOT Act and all things Executive Order 9066/Korematsu) that were likely disproportionate to the odds of the same or similar swan reoccurring, given all other external factors.

On an obviously much smaller scale, the same dynamic appears in basketball prospect analysis. At the time they were drafted, there were no players in the NBA that matched the skillsets of Steph Curry, Nikola Jokic, Kevin Durant, Giannis Antetokounmpo, etc., even as prospects. 

Throughout league history, players who reach that level of impact tend to do so by introducing something new, not by replicating what already exists. However, that hasn’t stopped teams from making decisions and allocating resources in an attempt to chase prospects that vaguely resemble current stars in the hopes of re-catching lighting in a bottle.

Unique Production

My issue with this type of analysis extends beyond black swans. Much of the stats-driven community’s current modus operandi when it comes to prospect evaluation is “Minimum Production”: using historical precedent to group prospects based on statistical baselines as a means of projecting future success.

Minimum Production analysis is a great starting point, one that I have often relied on, and can help a beginning evaluator determine what they should and should not care about, but it misses the ultimate point.

Just like the original American prospecting (the 1840s gold rush), the first mover advantage in the draft is real and those who follow what’s last will go broke missing out on what’s next. Our view of what superstars should look like is often shaped by the current landscape and composition of the league’s best players. This deductive approach to player analysis fails to account for the fact that the league’s most successful players achieve that success by bringing entirely new combinations of skill sets into the league, not by conforming to existing molds of greatness.

Instead of trying to pattern-match prospects to past successes, we should be trying to identify players who win in their own unique ways, whether through a rare combination of skills or through an unprecedented magnitude of a single skill (“Unique Production”). The reason people struggle to think of a well-fitting, one-to-one player comparison for Cameron Boozer isn’t the reason he’s going to struggle at the next level; it’s what’s going to make him thrive. Find the next black swan instead of recreating the last one.

Upside > Everything: The Role Player Problem

A natural counterargument to the Unique Production framework is that it applies primarily to identifying stars, and that well-rounded archetypes still have value when drafting for role players. A common rationale for prioritizing “high-floor” prospects follows from this thinking: safer prospects are more likely to become useful NBA players. The implicit assumption is that avoiding failure has meaningful value. However, this assumption breaks down when you consider how few players actually matter and how sharply player value scales in the current NBA.

Pillar 1: most players don’t matter. 

On its face, the biggest problem with drafting for a player with only “role player” upside is that most role players are highly replaceable. A common pro-expansion argument you hear is that the league is so talented that good players aren’t able to see the court in the current 30 team league. Whether you believe in that or not, the replacement level player is really good, and only getting better by the year. @sradjoker did a study concluding that only 30% of NBA players end up having positive impacts on their team. The threshold for being good is high.

A team might hit on a “role player”, but that player could still be one of the role players that isn’t good enough to drive winning in a meaningful way over the replacement. This is especially true in the playoffs. @ChuckingDarts had a great tweet on the subject: when it comes to players that actually make a difference and can get on the court in a playoff series, even the non-stars reach incredibly lofty heights. The further you get into the playoffs, there are no role players, only the elite and the replaceable. Your favorite "serviceable rotation player’s” job can be done by somebody else for cheaper or by better players through more optimized playoff-like rotations. 

In a league where only a small fraction of players meaningfully impact winning, drafting with limited upside in mind all but guarantees that you miss the outcomes that matter.

Pillar 2: the value of impact is exponential.

Not only is the number of players that actually create a meaningful difference, with respect to winning games over the replacement level player exceedingly low, but the value of players increases exponentially (e.g., the jump from 90th percentile NBA player to 95th percentile is more valuable than the jump from 50th percentile to 70th percentile), making a small number of players responsible for a disproportionate share of winning.^[1]

Because of this, the “best” players are so much more impactful than the merely “good” players that a outsized share of league value is concentrated among a relatively small number of outcomes. Targeting players solely by how likely it is that they can reach these outcomes is a more than viable strategy. The best part of this strategy, despite how it’s often framed, is that it doesn’t forfeit access to the desirable middle outcomes. For the most part, it means drafting a player whose outcome distribution still contains the very same role-player outcomes you're hoping to secure from the "safe" option, while also retaining access to outcomes the safe option can never realistically achieve. The high-upside prospect offers a chance at both the role-player outcome and something much better.

It’s also important to qualify that I’m not opining on which archetypes of prospect are high ceiling or high floor, my only assertion is that you should place a premium on whatever high ceiling/upside looks like to you. Last draft cycle, there were reasonable arguments that Collin Murray-Boyles was both a high ceiling player and a low ceiling, high floor player. I’m not here to lecture you about which school of thought is correct (high ceiling, obviously), but I am saying that if you viewed him as nothing more than a safe, rotation bet, then it would be suboptimal – and maybe even antithetical to the exercise as a whole – to rank him above any prospect that you believe has even a chance to make an All-Star game.

Visual Representation of Pillars 1 and 2

The best way for me to visualize my thinking is to pretend this is Econ 101, so let’s suppose, hypothetically, that we can perfectly predict a player’s outcome distributions and all players can be represented by the below bell curves — with a normal curve representing the average draft pick, a slightly left skewed curve being a “safe, high-floor player”, and a flatter, more evenly distributed among outcomes curve, to represent “high ceiling, low floor” picks. The x axis shows the range of value – however you may define it– of all draft picks and the y axis is the probability of a player reaching a given value.

This graph on its own doesn’t really tell us anything, it merely depicts how we might conceptualize floor and ceiling outcomes. The question is whether one profile is preferable to the other, and when we update the graph to reflect the two aforementioned pillars, the takeaways become clearer. 

First, we remove all player outcomes that fall below our arbitrary replacement value line, since those players’ impact is almost perfectly elastic given the current environment of the NBA.^[2]

Second, for comparison purposes, we remove the median outcome curve. Then, for both the high floor and high ceiling curves, shade in the area underneath its curve but above the other profile’s curve. This area, often called surplus, represents the excess value of each bell curve in comparison to the other.

The green area represents outcomes uniquely captured by the high-upside profile. The red area represents outcomes uniquely captured by the high-floor profile.

Even after removing the below replacement level players, a pure surplus comparison (green area vs. red area in graph 3) doesn’t provide clarity as to why high upside drafting is superior. Depending on how you draw the curves and where you draw the replacement-level line, the area under either curve could be larger than the other. 

In other words, Pillar 1 replacement-level considerations alone are insufficient to justify drafting for upside. A world where low-end outcomes are replaceable is not enough to proclaim that high-risk profiles should always be preferred. To arrive at an “upside > everything” conclusion, we must also account for Pillar 2 and the exponential effect of impact – and once we do, the picture changes dramatically.

The below is how I imagine the graph would look if you multiplied the above surpluses by the scaled effect of impact discussed in Pillar 2.

After weighting outcomes by their impact on winning, the green surplus expands dramatically, showcasing the disproportionate value of the highest-end outcomes.

To reiterate, none of this is an argument that any particular archetype is inherently high ceiling or high floor. It’s merely an argument that, whatever upside looks like in your evaluation framework, it should be aggressively pursued. In a league where replacement-level outcomes are abundant and superstar outcomes are disproportionately valuable, avoiding failure is not a strategy.

Reducing Variance

Waxing poetic about how valuable theoretical upside is begs the question of how I define upside. Perhaps counterintuitively, the players with the most long-term upside are the ones whose skillset most reduces short-term (game-to-game, possession-to-possession) variance. Basketball games are often decided by luck. Hot or cold shooting stretches, injuries, and what mood the ref is in often help determine the outcome of games. Randomness plays a huge part of basketball, but the players with the highest ceilings are often the players who exert the most control over variance itself. Superstars are typically the players that can suppress basketball’s natural volatility possession after possession, imposing their will on their team, the game, and the outcome.

An important distinction to make is that reducing variance in performance quality, often discussed in terms of consistency, is not being discussed here. Rather, the value is in reducing the effects of randomness, or luck, within a basketball game. Think of the worst pickup basketball game you’ve ever watched or been part of. Both teams are filled with equally talentless players chucking up as many threes as they can hoping that their team will get luckier bounces than the other. Many cynics view the modern NBA as a more skilled version of this same pickup game – a contest so optimized that it comes down to who gets lucky on any given night. That’s where the casual fan gets it wrong. The best players and teams are creating their own luck by reducing the elements of chance through skill. Players can create their own luck in three primary ways:

1. Rimfluence

What started as the title of my foray into creating RAPM metrics evolved into the term that most centrally guides what skills I value in player evaluation. The premise is simple: paint dominance wins basketball games. Whether through at-the-rim scoring, rim protection, interior playmaking, or controlling the glass, Rimfluence is how much a player impacts the game in the most important area of the court.

The rim is the lowest-variance area in basketball. Layups are the highest percentage shot one can take, and also have the highest probability of leading to (i) another opportunity to score even if the original shot is missed – through offensive rebounds – and (ii) free throws, the only shot in basketball where the opponent isn’t allowed to play defense (not to mention the secondary benefits of getting opposing players in foul trouble and one’s team in the bonus). Players who consistently create or deny rim attempts shift possessions toward predictable outcomes for their team. Rimfluence compresses randomness, making every possession less dependent on shooting luck and more dependent on repeatable advantages.

NBA examples: Victor Wembanyama, Giannis Antetokounmpo

2. Efficient Creation

In a team game, self-creation is the ability to generate efficient offense independent of circumstance. Engines can generate offense in the absence of, and oftentimes after the destruction of, structure. Great creators can reliably manufacture quality shots regardless of external factors, preventing possessions from devolving into coin flips. The ability of a player to drive offense no matter the quality of teammates surrounding them or type of defensive coverage in front of them is what separates sustainable offense from situational offense. The fewer variables required for a player to produce, or help their teammates produce, the less room there is for randomness to determine the outcome.

NBA examples: James Harden, Kevin Durant

3. Possession Economy

In general, the number one way to reduce the effects of variance is to increase the sample size. In basketball, the sample size is the number of possessions. The more possessions a team controls, the less each individual possession matters – randomness loses influence as the sample grows. When your opportunities increase at the expense of your opponent's, success requires less perfection. Possession creators are the insurance policy against imperfect performance and bad luck.

Growing the possession economy comes in two forms, creating more possessions for your team and ending the other team’s possessions early. In regard to the former, there are both active and passive means of participation. The primary way for a player to actively participate in possession creation is offensive rebounding. These extra opportunities, usually resulting in layups or open threes, extract immense value from possessions that otherwise wouldn’t exist. In an eating competition, the person who gets to take the most bites will usually win, so naturally, teams should want players who give them more bites at the proverbial apple.

Limiting turnovers is the passive way to impact the possession economy. Its' impact typically goes unnoticed, only revealing itself as a relative advantage that slowly accumulates over the course of a game. Consistently preserving opportunities can be just as important as occasionally creating them. If offensive rebounding gives you more bites at the apple, limiting turnovers makes sure you don’t drop the apple before it gets in your mouth.

Conversely, suppressing the number of opponent possessions, or more accurately, the number of opponent possessions that end in a shot attempt, is the final way to influence the possession battle. Generating steals and offensive fouls remove opportunities for the opponent to get lucky, while often vaulting the defensive team into transition situations, the possession type that most often leads to points.

All 500+ NBA players are supremely talented at the things you can work on in an empty gym, which is why the real value lies in the players who allow their teammates to showcase those talents more often.

NBA examples: Ausar Thompson, Dyson Daniels

2026 NBA DRAFT BOARD

If you’ve made it this far, you know all the stats, have seen all the queries, and understand the ins and outs of this class. For that reason, I’m forgoing traditional draft notes and opting for the below graphic, which applies the Unique Production and Variance Production framework to the top 20 players on my 2026 draft board to give you a general feel for how I think of each player's skillset.

Epilogue: The Purpose of the Space – Knowledge Decay and the di Lampedusa Principle

In a space where countless hours are devoted to a hobby that has no real-life “payoff”, it seems worthwhile to occasionally reflect on the purpose of the exercise and what role the people who participate in it play.

The thing that’s always bothered me most about this space is its obsession with right and wrong and inability to find value in ideas/processes that produce unpopular results. Of course, the ultimate goal is to be “right” in your evaluations of players, but given that winners and losers of a draft can't be properly litigated for years and years, there has to be a secondary force driving the passion to continue devoting time and energy to this hobby. For me, that secondary force is the excitement of creating something new and interacting with people who think and discuss my favorite sport in ways that I haven’t previously considered. There are no stakes in this hobby, there’s nothing to gain from parroting the established, top-end analysts. The goal should be to discuss the sport in an interesting way and, hopefully, change what top-end analysis looks like. The pursuit of a new idea should be far more rewarding to me than the constant reapplication of something previous in the hopes of maybe having a good "Tawny Park Ranking."

The good news is that the pursuit of new and the pursuit of correctness go hand-in-hand.

In our draft world there’s nothing on the line, no games are won or lost, no money is earned. The real currency and the real value is the quality of one’s ideas. And like a new car, the moment you drive that idea off the lot and post it onto the internet, its value as a means to beat the competition is drastically slashed. Once an insight turns into common knowledge, it becomes an assumption.

There’s a famous Henry Ford quote that goes along the lines of “if you do what you’ve always done, you’ll get what you’ve always got.” As it relates to the draft space and the relative quality of one’s work, I couldn’t disagree more. In player evaluation, the target is always moving. Knowledge, or at least the effectiveness of knowledge, decays. It doesn’t even have to become incorrect to become less valuable. Often it simply becomes common knowledge. The first person to recognize the importance of age, efficiency, impact, or any given indicator gained a meaningful edge over the field. The thousandth person gained no such advantage. 

Further, the way in which we evaluate success can shift entirely, rendering previously useful methods of analysis ineffective. In 2006, the International Astronomical Union changed its definition of a planet. Overnight, Pluto went from being our solar system’s ninth planet to being its first modern former planet. Did Pluto itself change, or did we change how we evaluated Pluto? Was the scientist that originally determined Pluto to be a planet wrong? No. Would that same scientist be wrong to declare Pluto a planet today? Of course. Pluto still didn't change, the framework used to evaluate Pluto changed. Draft analysis is no different.

The goal of draft analysis has never changed, but the methods used to achieve that goal, and the information available to evaluators, constantly do. Because of knowledge decay and the comparative nature of successful prospect evaluation, I'm throwing out Henry Ford's line of thinking in favor of the di Lampedusa principle: everything must change in order for nothing to change.

What allowed one to outperform analysts whose primary focus was points, rebounds, and assists is completely different than what’s required to outperform those who index for Box Plus-Minus, VORP, RAPM, EPM, etc. Many of the basketball truths we currently hold to be self-evident will be relics of a past era in short order, much like how raw field goal percentage or points per game are viewed today. By simply continuing to do what was successful yesterday, you're falling behind today, and even more-so tomorrow.

None of this is to say results don't matter. The NBA is a results-based business after all. But in this space, results arrive years after the decisions that produced them and by the time an evaluation can be properly judged, the conversation that led to that analysis is stale.

I leave you with the same advice that I started this article with: find what's next instead of recreating what was last — you'll definitely have more fun.