That leveraged single-stock exchange-traded funds will eventually maul the retail investors who love them is, by now, close to consensus.
My Bloomberg Opinion colleague Shuli Ren made the case last month that investors in leveraged ETFs tied to semiconductor stocks such as South Korea’s SK Hynix Inc. are at particular risk of mauling. The cyclicality of the memory chip industry is a broad concern; but those who’ve enjoyed the supersized gains of leveraged ETFs must also worry about the inescapable math of these products, called volatility drag, which grinds down the funds in choppy markets.
The questions Ren left open are the ones that matter next: Does the damage stay with the people holding them? And if not, are the banks and money managers involved in structuring these products next in line?
The buyer of a 2x single-stock ETF is not a counterparty to this trade; he is the raw material. His return is a loan from the future that the mechanics of the product are built to call. A stock levered 2x daily can only keep rising forever, which it won’t; fall, in which case leverage turns a 30% correction into a catastrophe; or chop sideways, in which case volatility drag quietly evaporates the fund. Two roads lead to ruin, and the third doesn’t exist.
Now follow the money out of the investor’s pocket. Some of it vanishes: Volatility drag is not a fee, and no one pockets it — it is the arithmetic cost of resetting leverage along a bumpy path.
The rest is very much in someone’s pocket. There are fees and financing spreads on the swaps that manufacture the leverage. There is insurance: The banks issuing total-return swaps to the ETF sponsors buy protection against the underlying stocks collapsing. They do this largely through cliquets — chains of options that reset to the current price at intervals, paying out leg after leg as a stock falls; their cost has more than tripled for the hottest names over three months. And there is opportunistic trading: The end-of-day ETF rebalancing is large and predictable, so market makers front-run it and skim the difference. The gap between what these ETFs return and what a friction-free version would deliver is the toll, and it is rising.
The banks are not holding the crash; they bought protection. The trouble, if it comes, will land on whoever sold it to them.
The cliquets and swaps are absorbed by yield-reaching asset managers and hedge funds, who collect large premiums in exchange for wearing the tail risk. They are short a book of correlated AI gap risk. The fatter the premium grows, the more of them are tempted in. This is the next group in line after retail — the carry-hungry institutions that wrote the crash insurance.
It is the oldest structure in finance, the one that detonated in the 2018 volatility blowup and, in a grander key, in 2008, when monoline insurers who sold cheap insurance on mortgage-backed securities were vaporized when the tail risk arrived. Someone is always short the tail; usually whoever needs the yield.
Could an AI crash reach the big banks? No single name is large enough, the derivatives are disclosed and the resets are daily. But Bloomberg Intelligence counts more than 450 leveraged and inverse single-security ETFs launched since 2022, a category past $150 billion and stacked overwhelmingly on a tightly correlated cluster of AI and semiconductor names. Roughly a quarter of all new US ETF filings are now leveraged or inverse.
These products do not fail one at a time, because they share a fuse: the same outlook for AI capital spending that lifted all of them at once. When the trend breaks, they rebalance in the same direction, into the same closing auction, on the same afternoon.
Nomura Holdings Inc. estimates leveraged ETFs already generate some $9 billion of rebalancing demand for every 1% market move; Barclays Plc. puts recent US rebalancing at several times its long-run average. That is not one exploding cigar. It is a box of them wired to a single fuse.
South Korea’s top financial regulator has openly regretted approving 16 leveraged chipmaker ETFs in May, saying the negative side-effects of the products have grown significantly. Europe offered the cleaner warning: A 3x short AMD product there was wiped out in a single day’s rally.
The US does have a guardrail that has capped leverage in these products. Under Rule 18f-4, the Securities and Exchange Commission anchors a leveraged fund’s risk test to the asset it tracks, which is why every 3x and 5x single-stock application has been rejected, and the ceiling holds at 2x.
But that leaves open the danger of 450 of them, every one individually compliant, every one pointed at roughly the same trade, selling together. And the correlated pile is about to grow: SK Hynix lists on Nasdaq on July 10, and the moment a hot AI name trades in New York, the issuers race to lever it.
The system is stable for now in the way a row of dominoes is stable: every piece individually hedged, the eventual loss already sold — to the retail buyer who mistakes a paper gain for profit and to the yield-hungry fund that wrote the crash insurance. The only open variable is how tall the pile grows before the AI trend finally breaks.
A message from Advisor Perspectives and VettaFi: Discover something new! Click here to register for our upcoming webcasts.
Bloomberg News provided this article. For more articles like this please visit
bloomberg.com.
Read more articles by Aaron Brown
