138 | Daryl Morey on Analytics, Psychology, and Basketball
What this covers
Daryl Morey, general manager of the Philadelphia 76ers and former architect of the Houston Rockets' analytics revolution, joins Sean Carroll to discuss how disciplined decision science—the systematic application of data, behavioral economics, and cognitive bias research—compounds into sustained competitive advantage in basketball. The conversation treats the sport as a laboratory for understanding optimization and human psychology more broadly. Morey argues that a consistent 3–5% edge, built from better decisions across draft, trade, and free agency, compounds over time into championship-level performance. Rather than replacing human judgment, data grounds it: scouts remain essential, but their intuitions are sharpened when forced to account for what the analytics consensus reveals.
The discussion ranges across several distinct territories. On combining data and human judgment, Morey details how teams can counteract cognitive distortions—the endowment effect (overvaluing your own assets) and anchoring—by deliberately inverting trades to see them from the other side. On strategy, he argues that intuitive game-theory objections to feeding your best player prove overblown when subjected to rigorous positional analysis. On draft evaluation, he finds that interviews add almost no predictive value over track record, and that reward function tuning is necessary to avoid models that merely predict "nothing happens." On broader sport culture, he observes that emotionally acceptable and unacceptable ways to lose explain why analytically sound strategies face resistance. The pair also diverge on dark matter—Carroll defends it as the simplest model unifying independent phenomena; Morey expresses skepticism rooted in historical epistemology. Throughout, the focus remains on how institutions can think more clearly under uncertainty.
Morey argues that decision-making in basketball is a tractable optimization problem won by disciplined decision science—combining data with human judgment, correcting for cognitive biases, and chasing small compounding edges—rather than by intuition or machismo alone.
- A consistent 3-5% edge from decision science compounds into long-term winning
- Human biases (endowment, anchoring, confirmation) systematically distort sports decisions and must be actively countered
- Data grounds human intuition rather than replacing it; both scouts and analytics are needed
WIMP candidates warrant reduced credence from null experimental results, but dark matter itself remains robust.
- The weakly interacting massive particle was the favored dark matter candidate because such a particle naturally arises and would have the right cosmic abundance, but since experiments had a roughly 60% chance of detecting it by now and have not, credence in the WIMP specifically should fall—though this barely dents confidence in dark matter, which could instead be a harder-to-detect candidate like the axion.
“we could have found it by now... the chances that we would have found it by now, are maybe like 60%... So at some point, your credences have gotta go lower”
- The cancellation of the Superconducting Super Collider in Texas—which would have been more powerful than and earlier than the Large Hadron Collider—reflects the US entirely abandoning the attempt to build world-class particle accelerators, leaving the next major collider to be a contest between China and Europe.
“The United States has entirely abandoned even the attempt to build world class particle accelerators. So right now, the debate on what the next one will be is between China and Europe.”
Dark matter simultaneously explains multiple independent phenomena with remarkable predictive success.
- Unlike a single ad-hoc gap-filler, dark matter is one very simple model that simultaneously explains a dozen independent phenomena—galaxy rotation curves, cluster dynamics, the cosmic microwave background, and structure growth—and crucially the CMB came out exactly as dark matter predicted in advance, which is almost impossible to reproduce by modifying gravity alone.
“the thing about dark matter is we have a dozen holes and this one thing fills them all, whether it's the rotation curves of galaxies or the dynamics of clusters or the cosmic microwave background or the growth of structure, one very, very simple model fits all of it”
Scientific fields benefit from practitioners holding both advocacy and critical stances toward their theories.
- Scientists split between those who see their job as being their theory's biggest advocate and those who see it as being its harshest critic; both attitudes are valuable at different times, and there is no single right answer, so a healthy field benefits from people of both dispositions.
“I know great physicists who will articulate the idea that it is their job to be their theories' biggest advocates. And there are other people who think it's their job to be their theories biggest critics”
Emotional attachment to ideas, though biasing, may be necessary for individuals to pursue major discoveries.
- The same emotional attachment to ideas that biases people may be necessary for major discoveries, since you almost need obsessive individuals working against their own interest to push breakthroughs forward.
“it's also what drives them to make great discoveries... You almost need obsessive people doing things that are against their own interest to create some of the great discoveries.”