by Editorial Team, AdvisorAnalyst
The question Pierre Daillie puts to Larry Swedroe near the top of the conversation1 is direct: is AI making markets easier or harder to beat? Swedroe's answer is equally direct, and it sets the tone for everything that follows.
"In the very short term, AI makes markets a little more beatable," he says, "but that makes the market more efficient, in and of itself." This is Andrew Lo's (MIT, AlphaSimplex) adaptive markets hypothesis in its most practical form — inefficiencies attract capital, capital corrects pricing, the anomaly shrinks. The process is self-defeating for followers, and increasingly dominated by world-class institutions with the compute power and incentives to find every remaining penny.
Swedroe walks the room through the full arc. Before the CAPM, there were no finance programs in American universities. The single-factor model explained roughly two-thirds of portfolio variance — leaving one-third on the table for anyone paying attention. Fama and French's three-factor model pushed explanatory power to 93 percent. Carhart added momentum. Novy-Marx added profitability. By the time AQR's team had incorporated quality, the five-factor model explained 98 percent of cross-sectional returns — and Warren Buffett's alpha, which had survived decades of scrutiny, all but disappeared. "He hasn't outperformed the market in the last 17 years or so," Swedroe observes. Not a failure of genius, but a consequence of markets catching up.
The AI chapter in this story is continuous, not new. Renaissance Technology and Citadel have deployed machine learning for decades, Swedroe notes, "extracting tens of billions of dollars from the market by finding micro-inefficiency." What has changed is speed and sophistication. Earnings calls that once took hours to reprice are now repriced in seconds. Nonlinear relationships invisible to regression are now surfaced routinely. But the same mechanism that rewards early movers eliminates the opportunity: when AI discovers an anomaly, money flows in, prices adjust, and the premium compresses. The danger, Swedroe is careful to add, runs in both directions. "AI tools are great at data mining. They'll find a correlation, but it may not mean anything." His standard for a durable factor — persistent, pervasive, robust to definition, implementable, and grounded in either a risk or behavioural explanation — applies equally to machine-generated signals. "Without that, the odds are good that what you have found is a data mining outcome," he says, citing the notorious example: a researcher once found that the best statistical predictor of the S&P 500 was butter production in Bangladesh.
From efficiency, Swedroe moves to diversification — and here his message was blunter. The conventional 60/40 portfolio is not what most investors believe it to be. "Most people, if you ask them how much of their risk in a million-dollar portfolio with $600,000 in equities is in equities, and they say 60%. And I point out that's wrong — it's probably closer to 90, because the equities are much riskier." That single correction reframes the entire planning conversation. If investors believe they are balanced when they are not, they are carrying tail risk they have not measured, cannot articulate, and will almost certainly fail to endure when tested.
The solution is what Swedroe calls hyper-diversification: genuine exposure to reinsurance, private real estate, long-short factor strategies, private credit, and trend-following — assets with empirical evidence of premium persistence and, critically, near-zero correlation to equity and bond markets. "There's nothing more logical than investing in reinsurance, literally nothing," he says, "because there is no reason to believe that hurricanes or earthquakes cause bear markets or vice versa." He holds 50 percent of his own portfolio in illiquid alternatives. He was present through multi-year drawdowns in both reinsurance and AQR's long-short strategy — and watched most investors exit just before the recovery. The reinsurance fund he references returned 146 percent over three years. Two-thirds of investors were gone by then.
The behavioral problem is the portfolio problem, and Swedroe is unsparing about it. "Investors like to think they're financially rational, but they're psychologically rational. They end up making mistakes because their stomach is making decisions, not their head. And I've never met a stomach yet that makes good decisions." The practical test he applies is deliberately visceral: tell clients their $600,000 equity position just fell 60 percent, $360,000 is gone, and you now need them to buy more. "Are you going to be willing to do it? And don't lie to yourself for me." The exercise surfaces behavioural truth that abstract risk questionnaires never reach. Most advisors, in Swedroe's estimation, never run it.
Mike Philbrick's return stacking framework earned Swedroe's direct endorsement — qualified precisely where it matters. Two non-correlated 10-volatility assets don't produce 20 volatility; they produce roughly 14. The math is sound. But "you could get uncorrelated assets correlating at the wrong time," Swedroe cautioned, "and now your volatility isn't even 20 — it could be 30." Planning must account for worst-case correlation, not average-case correlation.
The forward prescription Swedroe offered advisors was unambiguous. Pure asset management, he says, has no future. "I don't think those people really have a chance to exist much longer." What advisors must become are genuine wealth managers — integrating investment, estate, tax, insurance, and family governance into a coherent plan. The second obligation is competency in alternatives: understanding the instruments, performing real due diligence, and educating clients on the actual trade-off. "I believe a large part of the investor community is not being exposed to these alternatives," he says. That is not an access problem. It is a planning failure with compounding consequences. Monte Carlo analysis, run properly, translates the cost of avoidance into explicit probability: a 15 percent chance of portfolio failure on a conventional 60/40 at 4 percent withdrawal drops to 4 percent when uncorrelated alternatives are included. "You're trading off illiquidity risk, which you say you don't like, but your chance of failure fell in relative terms like 75 percent." Advisors who cannot make that argument will lose the clients who need it most.
Five Key Takeaways for Advisors and Investors
1. AI accelerates efficiency — it does not create durable alpha for followers. The institutions finding micro-inefficiencies are closing them as fast as they find them. For most investors, AI-adjacent investment products warrant the same scrutiny as any other factor: is the premium persistent, pervasive, robust, implementable, and logically explained? If not, it is data mining dressed as innovation.
2. The 60/40 portfolio is 90 percent equity risk, not 60. True diversification requires assets structurally uncorrelated with stocks and bonds — reinsurance, private credit, long-short factor strategies, trend-following. Owning all of these at modest allocations cuts left-tail risk in ways that no overweight to any single asset class can replicate.
3. Behavioural risk dwarfs market risk for most investors. The decision to sell in a drawdown and the subsequent failure to re-enter are the actual destroyers of wealth. Advisors must stress-test risk tolerance with absolute dollar numbers — real losses on real portfolio sizes — not percentage abstractions. The stomach is not a risk manager.
4. Self-healing assets reward only those who stay. Reinsurance, AQR-style factor strategies, and trend-following have all endured extended multi-year underperformance before delivering outsized recoveries. The investors who captured those recoveries are precisely those who did not exit. Sizing allocations appropriately — so underperformance is tolerable, not catastrophic — is what makes staying possible.
5. The advisory practice of the next decade is built on wealth management and alternatives literacy. Asset management alone is a diminishing business. Advisors who can run Monte Carlo simulations, explain illiquidity trade-offs in probabilistic terms, evaluate alternative vehicles with genuine due diligence, and integrate those recommendations into a full wealth plan will define the profession going forward. Those who cannot will be displaced by those who can.
Larry Swedroe is the author of 18 books on evidence-based investing and currently serves as an outsourced CIO and consultant to registered investment advisory firms. This episode aired on Raise Your Average, hosted by Pierre Daillie of AdvisorAnalyst.com and Mike Philbrick of ReSolve Asset Management.
Footnote:
1 "Larry Swedroe: The Adaptive Market & The Undiversified Investor." AdvisorAnalyst, 22 May. 2026.
