Passive flows have settled the question of which stocks to own. They have not touched the question of when to own them.
Bloomberg’s Justina Lee reported on August 19, 2026, about a paper that hands one of investing’s oldest arguments something close to a smoking gun. Hannah Unterberg of UC Irvine finds that at the start of each month, when retirement contributions and other mechanical allocations send predictable bursts of cash into passive funds, the stock pickers who stray furthest from the benchmark fare measurably worse — nearly five basis points a day when flows are strong. Passive vehicles now account for 63% of all equity funds, against 42% a decade ago; last year, per S&P Global, 21% of active U.S. large-cap funds beat the S&P 500. Unterberg estimates that active alpha before fees has fallen to roughly −0.66% a year since 2010, and that active share — once a predictor of outperformance — now predicts the opposite.
Terry Smith supplied the operator’s version in his July letter to investors: “As active fund performance continues to worsen, more people abandon it, producing a pernicious feedback loop… a buy and hold strategy can only work if you are not subject to flows, and we are.”
The conventional inference is to surrender and buy the index. That inference is exactly half right. The missing half is the entire investment case for what we do.
Two decisions, not one
Equity investing contains two decisions, and they are not the same decision. The first is cross-sectional: which securities to own, and in what weights. The second is time-series: how much market exposure to carry, and when. Virtually every word written about the decline of active management concerns the first. Almost none concerns the second.
Indexing answers the first question elegantly and the second not at all. The passive investor has outsourced security selection to the market’s own capitalization weights and has, by construction, accepted full exposure at every moment — through calm, through crisis, through the worst ten sessions of the decade.
The Demeter Dual-Engine Strategy takes the opposite division of labor. We make no security selection decision whatsoever. Engine One is levered long index exposure, implemented in S&P 500 and NASDAQ futures. Engine Two is fully liquid cash with zero equity market exposure. A daily, rules-based signal determines which engine is running. We don’t pick stocks. We pick when.
Sharpe’s arithmetic binds one game, not the other
Lee’s piece notes the simplest explanation of all, supplied by William Sharpe in 1991: before costs, active stock picking is zero-sum, because active investors collectively own the market; after costs, they must do worse. That is arithmetic, not theory, and cannot be argued with.
But notice what it constrains. Sharpe’s identity governs how a fixed pool of shares is distributed among owners. If I overweight a name, someone must underweight it. It is airtight about relative weights inside the pool.
It says nothing about whether to be in the pool. When our signal moves the strategy to cash, we are not taking a dollar from another stock picker. We are declining to bear equity risk through a window in which the compensation for bearing it is poor, and transferring that risk to a counterparty who wants it. Futures are a risk-transfer market, not a zero-sum information contest, and the payment for standing on the other side of institutional risk aversion — the volatility risk premium — is among the best-documented phenomena in financial economics.
Sharpe’s arithmetic tells you that changing lanes on a crowded highway is zero-sum: every car you pass is a car that falls behind. It tells you nothing whatsoever about whether to drive during the blizzard.
Informational alpha decays. Structural alpha does not.
Why should any systematic edge survive when value and momentum did not?
Because there are two species of alpha with very different half-lives. Informational alpha is the return to knowing something others do not, or processing it faster. It arbitrages away by construction: once a signal is widely known, capital bids the opportunity away. Cross-sectional factor investing is largely the history of that decay. Fama and French published value in 1992; it worked for two decades before compressing after 2010. Jegadeesh and Titman’s momentum followed a similar path.
Structural alpha is the return to willingly holding a position others are structurally, regulatorily, or psychologically unable to hold. Publication does not compress it, because publication does not remove the constraints that create it. Institutional option sellers still demand payment for bearing left-tail insurance risk, because their risk committees, capital rules, and ordinary loss aversion penalize tail losses far more than they reward tail gains. Carr and Wu published the variance risk premium. It is still there.
The five academic phenomena the Dual-Engine signal harnesses — volatility clustering, the leverage effect, mean reversion, regime switching, and the volatility risk premium — are structural, every one of them. The strategy does not require the market to misprice anything. It requires markets to keep behaving asymmetrically around stress, a far more durable precondition than cross-sectional mispricing.
Which is what makes Unterberg’s paper interesting to us rather than threatening. Her claim is that who is buying, how much, and who is left to sell to them moves prices — and that passive-flow pressure is larger and longer-lasting, while active-flow pressure reverses. That describes a market in which price is increasingly set by mechanical, non-informational flow. Smith’s complaint about momentum displacing fundamentals, and about 33% daily moves in large stocks, describes the same market with precision: he is describing volatility clustering and leverage-effect asymmetry, while running a mandate that cannot act on either. The force degrading cross-sectional edge is intensifying the time-series conditions our signal was built to read.
The algorithmic objection resolves the same way. The machines executing most equity trades were coded by the humans who preceded them, and they encode human risk frameworks rather than erasing them: risk-parity books deleveraging into a volatility spike, pension programs selling risk into drawdown. Automation has institutionalized the asymmetries, not eliminated them.
Rolling bubbles: the index does the wave rotation for you
Macquarie’s global strategist, in the note circulating this summer, argues that we inhabit a world of “rolling bubbles” rather than one bubble waiting to deflate. Because AI is a general-purpose technology reaching escape velocity, the collapse of one bubble — software — creates the next, in LLMs, which fuels infrastructure, then robotics and biotech. His conclusion: “Identifying new waves while avoiding deflating winners of previous cycles will remain the key to successful investment.”
That is a brutal assignment for a stock picker and a trivial one for a capitalization-weighted index. Deflating winners shrink their own weight; rising waves grow into it. No sell decision is required, and nobody falls in love with last cycle’s champion. Owning the S&P 500 and the NASDAQ is a standing subscription to every wave — bought as it begins to matter, sold as it stops.
The same note argues that “wars might fester but economies and markets will find a way to navigate these uncertainties,” and that investors who overestimated Ukraine, Iran, or the Liberation Day tariff shock underperformed badly. Which deserves a plain statement for the journalists and regulators reading this: Demeter embeds no geopolitical forecast. We hold no view on Hormuz or on where an armistice line falls. The signal reads the tape; geopolitics enters only insofar as it registers there. That is why a world of more frequent shocks is not a threat to the architecture. Volatility is not our risk. It is our raw material.
What would falsify this
Candor is part of the argument. Our current external briefing discloses that 2026 year-to-date tape behavior sits outside the thirteen-year distribution against which the signal’s parameters were calibrated. A thermometer calibrated at sea level that reads strangely at altitude is not a broken thermometer — but that distinction must be tested, not asserted. So we monitor parameter-proximate diagnostics continuously, we pre-committed to structural-break conditions with a date attached, we track capacity daily through execution slippage, and we publish the analysis to allocators.
The passive revolution did not kill active management. It relocated it — out of the crowded, arithmetically constrained contest over which stocks to own, and into the far less crowded question of whether to own the market at all today. Two engines. One signal. Every day.
Jeffrey A. Sexton is Chief Investment Officer of Demeter Tactical Investments Corp.
Sources: Justina Lee, Bloomberg News, August 19, 2026; Hannah Unterberg, “Passive Flows, Active Woes” (working paper, not yet peer reviewed); Macquarie global strategy commentary, July 2026; “Five Phenomena: The Demeter Dual-Engine Quantitative Equity Strategy,” v21.1 redacted external edition, July 31, 2026. No performance data is presented in this article. Nothing herein is an offer to sell or a solicitation of an offer to buy any security, or investment advice. Past performance does not guarantee future results. For qualified investors and registered investment advisers only. © 2026 Demeter Tactical Investments Corp.
