Sources of Edge in the NBA Draft Market
Is the NBA Draft Market Efficient?
In theory, an efficient market is one where all available information is reflected accurately and instantly in prices, making it nearly impossible for any individual to consistently beat the market.1 This ideal arises naturally from the concept known as the “Wisdom of Crowds,” which suggests that collectively, diverse and independent groups of individuals can aggregate their private knowledge to arrive at remarkably accurate estimates and predictions—often surpassing the capabilities of individual experts. James Surowiecki’s famous example: a crowd at a county fair collectively guessed an ox’s weight almost perfectly when the average of guesses was taken – better than any individual guess.2
When applied to financial markets or sports betting markets, the strong form of this principle implies that betting odds or market prices should represent a distilled consensus reflecting all publicly and privately held information.3
Public consensus rankings for the NBA Draft function somewhat like markets, aggregating diverse evaluations into a collective judgment about player talent and future performance. Deferring to public consensus rankings relies fundamentally on the same underlying principle as efficient markets: that aggregating many independent judgments can yield more accurate predictions than relying solely on individual experts.
But unlike actual financial or betting markets, these rankings lack the direct financial incentives, immediate feedback, and rapid price discovery mechanisms that typically characterize efficient markets. And crucially, the Wisdom of Crowds only works if each person’s guess is informed by private information and made independently. Surowiecki identified four key criteria for a “wise crowd”: diversity of opinion, independence (avoid groupthink), decentralization (draw from local/private knowledge), and a reliable aggregation mechanism.
When these conditions break down, crowds become irrational mobs – as seen in speculative manias or bubbles. In a stock bubble, for example, all investors chasing the same hot asset create a self-reinforcing feedback loop divorced from fundamental value, until reality intervenes. For example, I really enjoyed reading Nate Silver’s recent piece about the recent failure of papal election prediction markets:
Nate’s subtitle summarizes the phenomenon: “Prediction markets and the conventional wisdom have limitations when they become feedback loops.”4
Likewise, an echo chamber of draft analysts afraid to stray too far from each other’s rankings can lead to a groupthink consensus. A diverse, independent set of evaluations would theoretically produce a better draft consensus but in practice, the draft media landscape can behave more like a herd than a collection of truly separate data points.
Draft analyst opinions are not independent – analysts read each other’s work. One or two respected voices (over the past several years, most notably ESPN’s Jonathan Givony) can heavily influence the consensus by seeding an expected rank order of prospects well in advance. There is signal in early rankings of prospects and Givony’s rankings have some amount of Wisdom of Crowds baked in, as he talks with scouts from a variety of NBA teams and weighs their own opinions alongside his own.5 But that far out uncertainty should be very high. And yet, as other draft analysts follow suit, a feedback loop, or a form of “informational cascade” in Surowiecki’s terminology, is created.
Without quick feedback on the accuracy of their predictions nor the hard accountability of losing money (or their jobs), public draft analysts are slower to correct course. In other words, the draft consensus behaves less like a wise crowd or efficient market and more like a hazy averaging of noisy interdependent signals. It bakes in widespread scouting knowledge, which has demonstrable value,6 but also groupthink and blind spots.
Sources of Edge in the NBA Draft Market
Agustin Lebron,7 quantitative trader and author of The Laws of Trading has an excellent perspective on finding edge in markets: “If you can't explain your edge in five minutes, you don't have a very good one.”8
SoBrief.com elaborates on the Edge chapter of Lebron’s book:
Edge defined. Edge is the reason why a trade makes money, the unique insight or advantage you possess that allows you to profit consistently. It's the story you tell yourself to justify taking a particular risk.
Simplicity is key. A good edge should be easily explainable. Complex, convoluted explanations are often a sign of overthinking or a lack of true understanding.
Given the dynamics at play in the NBA draft prospect ranking “market,” I propose a few different sources of edge:
Undo the Groupthink Before Applying the Wisdom of Crowds
Qualitatively9 net each individual draft analyst’s opinions relative to the Jonathan Givony anchor. From there, we can have a better sense of a draft analyst’s independent thoughts on a given prospect. Another way to go about this is to weight what the draft analyst is saying about a prospect’s skills (which is more often from first principles film evaluation) as opposed to the draft analyst’s ranking of the player (which is more influenced by other rankings).
Upweight “Superforecasters”10
Figuring out which analyst should be predicted to consistently beat the crowd estimate is an imperfect science without an exceptionally long track record (though some people, like the no-longer publicly posting Dean Demakis who goes by deanondraft11:
do have a long track record of predictions that add value above the consensus board. But here are some reasonable heuristics I take to sift through whose opinion is valuable:
1) Does the analyst understand what contributes to NBA value (more on this below)
2) Does the analyst at least take a cursory look at college player production metrics?12
3) Is the analyst early to recognize players that eventually rise?13
4) Does the analyst data mine / cherry-pick BartTorvik.com queries without logical reasoning behind those queries? An analyst from an NBA team told me he thought Torvik’s website’s player-season finder tool, while an excellent source of college basketball data, was the worst thing to happen to draft twitter14 since it makes it so easy to data mine and cherry pick data without thinking through causal mechanisms and without the accountability of out-of-sample testing.
Understand Shooting Variance
A player’s 3FG% can be a small noisy sample, and samples are much smaller than most people would like to admit for our desired level of precision. Incorporating additional information (FT%, 3FGA rate, 2 pt jumpshot FG%, shot difficulty, pre-college shooting data) to top-level 3FG% can provide a better prediction of how good of a jump shooter a player is at the next level.
Understand the Importance of Seemingly Small Age Differences
Using decimal age as opposed to integer age is a super simple change in information display and yet way more valuable than it seems. There is a large difference in one level of performance by a freshman whose age as of draft day is 19.1 vs. the same level of performance by a freshman whose age as of draft day is 19.8.
Understand the Predictive Power of Statistical Indicators
Strong pre-NBA steal rate, AST/TOV1516, rebounding rates, STL+BLK relative to fouls have all been shown in various draft models to be predictive of strong NBA performance (and lack thereof vice versa). Layne Vashro, noted NBA draft modeler and NBA champion analyst for the Denver Nuggets has previously stated:
Scorers, and shooters in particular, tend to be overrated. Steals seem to be underrated at every position. Guys who put up stats in the areas you don't immediately think about for that position (i.e. Blks/Rebs for guards or Asts for bigs) tend to be underrated, and guys who struggle in those areas tend to be overrated. Especially more recently, older players tend to be underrated.17
Similarly, after putting together his own draft model, Jacob Frankel stated that “in general, the same stats from position to position are correlated with success, not just the ones that would be traditionally associated with each position. Rebounding is really important for guards and assists are really important for centers.”18 NBA draft modeling has come a long way since the public models circa 2015, especially behind closed doors, but in general, most of Layne’s and Jacob’s takeaways should hold.
Understand Length > Height
In The Undoing Project, published in 2016, Michael Lewis wrote that in Daryl Morey’s early draft modeling days he discovered that “It didn’t matter so much how tall a player was as how high he could reach with his hands—his length rather than his height.”19 I could not find a direct quote from Layne, but back in 2015, Dean Demakis wrote that, “according to Layne Vashro, reach is more predictive than height, but height and wingspan in tandem are more predictive than reach.”20
I am shocked that it is 2025 and yet people still believe a player has good positional size based on their height, while ignoring contradictory length data.
I took wingspan and height data from CraftedNBA.com and ran a loess model2122 to model wingspan as a function of height, which gives the results in the chart below:
I chose to use wingspan for this example because 1) wingspan data was easily available from CraftedNBA.com, 2) it is an easy quick-and-dirty rule-of-thumb to say the average NBA player’s wingspan is about 4.5 inches longer than their height, and 3) prospects have no incentive to game their wingspan measurement.23
Understand Which Archetypes/Molds are More/Less Valuable at the NBA Level
One-Dimensional Scoring is Overrated
Microwave scoring without creation for others gets progressively more overrated as a player goes up in level, especially if the cost of expending energy on emptying a player’s bag is a corresponding loafing on defense. Let’s say a player can create his own 0.9 points per possession shot on demand given 18 or so seconds on the shot clock to empty his bag.24 When the alternative offensive action is to trust a bunch of high schoolers to create something without turning the ball over that 0.9 PPP on demand in the halfcourt can be insanely valuable. High school evaluators are not necessarily incorrect when rating “bucket-getters” very highly among high school players.
But the opportunity cost increases when the alternative is a college offense. 2024-25 D1 offensive efficiency was 1.062 PPP according to KenPom.25 And the opportunity cost is even greater when the alternative is an NBA offense. 2024-25 NBA offensive efficiency was 1.145 PPP according to Basketball Reference.26 Those numbers are global PPP that include transition and putback opportunities and, accordingly, overstate the efficiency of running halfcourt offense, but the principle and relative ordering holds. And to make matters worse, the level of coaching and advance scouting increases as a player goes up in level of competition. Better preparation to defend against one-dimensional players’ single dimension results in that dimension decreasing in value.
To be clear, shot creation is valuable, but it is especially so when it stresses the defense to defend both the player in question and the player’s teammates. But players like Cam Thomas, Zach LaVine, Julius Randle, and DeMar DeRozan, while incredibly talented, have skillsets much closer to “one-dimensional scorer” than “offensive hub.” And this shows up in their NBA impact metrics27 being underwhelming relative to their skill or talent level.
On-Ball Defense is Overrated and Off-Ball Defense is Underrated
Interestingly, this may be trending otherwise with the NBA increasingly becoming a weak-link sport.28 But historically, players like Avery Bradley and Klay Thompson, strong on-ball defenders who aren’t as consistent off the ball,29 have consistently shown up in defensive metrics as lagging their well-earned on-ball reputation. On the other hand, players like Robert Covington, weaker on-ball defenders who can rotate intelligently and create havoc with stunts and passing lane disruption30 have been underrated.
On-ball defensive ability (or lack thereof) is still important, especially at the lower end extreme, where 1) outlier small defenders without much length and/or 2) players who give little defensive effort31 can get relentlessly picked on. But in the bulk of the distribution, the most defensive value accrual occurs off-ball.
Applying the Edge
Perhaps arrogantly, I think that I can gain an edge relative to public consensus by applying the above principles, along with a dash of my own eye test32, to the 2025 NBA draft class. But even if I do not have the edge that I think I have, the process of putting my thoughts out in the open should clarify/sharpen my thinking and serve as a “decision journal”33 upon which to reflect and learn to improve my own process.
Between now and the 2025 NBA Draft I will write about some players who I believe are overrated or underrated by public consensus and my rationale for why that is the case. Stay tuned.
But in practice, Eugene Fama, the economist known for the Efficient Market Hypothesis, says the question is not as much “is the market efficient?” but rather “to whom is the market efficient?” See https://josephnoelwalker.com/eugene-fama-156/
I also enjoyed this prediction market trader’s explanation of how he went about trading this kind of market.
For example, this 2018 mock draft as of October 4, 2017 holds up remarkably well.
See https://cleaningtheglass.com/a-roll-of-the-dice-part-1/, but note that public consensus and actual draft order are not the exact same thing, especially further out from draft day.
Incredible surname
Or quantitatively – there could be some interesting Principle Component Analyses to be done here.
I.e. draft analysts who should be predicted to consistently beat the crowd estimate. See https://en.wikipedia.org/wiki/Superforecaster and https://www.amazon.com/Superforecasting-Science-Prediction-Philip-Tetlock/dp/0804136718
Also see https://deanondraft.com/ and https://x.com/deanondraft
I have my own college player metrics, which helps, but part of me thinks that you can explain a good chunk of the better vs. worse draft analyst dichotomy just by: who looks at college impact stats? And it doesn’t even have to be in an especially intelligent way… An analyst with an NBA team responded that it’s “absolutely maniacal how far BPM gets you” in draft analysis. This isn’t everything, as even the best available public metrics have some glaring flaws. And moreover, the way in which a player generates value at the college level may or may not reliably generalize to a higher level and/or to a different role.
For example, to my knowledge Sam Vecenie was the earliest public analyst to put Cedric Coward in the first round and Maxwell Baumbach was the earliest public analyst to put Coward and Yaxel Lendeborg on the 2025 draft radar.
I also enjoyed this Canzhi Ye tweet on the death of draft twitter.
Paradoxically, especially for non-PGs
Assists are a subjective stat subject to inflation/deflation and are not 1-to-1 correlates with passing skill. Nevertheless AST/TOV has been shown to have predictive power.
See https://en.wikipedia.org/wiki/Local_regression. You would also get similar results with a linear model, but the loess seems to model the extremes a bit better.
As this was quick-and-dirty I didn’t weight by minutes played, which, if anything, would increase the wingspan bar as better players tend to have longer wingspans relative to their height.
They may choose to alligator-arm their standing reach measurement in order to juice their vertical measurement.
For sake of argument, assume this estimate already bakes in offensive rebounding percentage.
DARKO DPM is the all-in-one metric based on the DARKO player projection system created by Kostya Medvedovsky. It is not a perfect metric, but it was voted as the best all-in-one metric in the public domain by a survey of team-side employees and basketball media.
Check out this excellent Thinking Basketball video:
Even low-effort defenders make most of their costly miscues off-ball, though, as players tend to take more pride in their more visible on-ball defensive opportunities and the average fan does not pay as close attention to off-ball miscues.
I try not to rely too heavily on this, though, as I don’t have evidence that my own eye test is any better than the average prospect evaluator.
As advocated by Sam Hinkie in his 76ers resignation letter
This is a banger
And you, my friend, are going to HATE Tre Johnson based on the criteria😂😂😂
Excellently written, and I wouldn’t have found with your re-post Jacob! Can’t wait to read your next piece.