The Whole Delta-Neutral Category Is Shrinking, and Almost None of It Is Depegging. The Stress Is Yield Compression, and It Applies to Us Too.

There is a quiet pattern across the synthetic-dollar market, and it is easy to miss because it does not look like a crisis. The category is contracting. Not depegging, contracting. Take the whole delta-neutral field as DefiLlama classifies it, forty protocols running some version of the basis trade, and add it up: it holds about $6.94B today against about $7.65B thirty days ago. That is a fall of roughly 9.4 percent, about $716M, in a month. Almost none of it is a depeg. Of the four largest dollars in the group, three sit within a tenth of a percent of a dollar today.

That combination has a name, and getting the name right matters, because the wrong name points you at the wrong worry. When a stablecoin's price collapses, it has depegged, and the story is about backing. When a stablecoin's supply collapses while the price holds, it has been redeemed or unwound at par, and the story is about demand. Almost everything shrinking in this category right now is the second kind. The stress the category is actually under is not peg failure. It is yield compression. And the honest thing for us to do, before we say anything else, is admit that it applies to our own yield too.

The category, at par, shrinking

Here is the field as public data shows it, read on the publication date. Every figure is recomputable from DefiLlama's stablecoin data, and we would rather you rebuild the table than take ours. The last two rows are in the table for contrast: they are not shrinking.

Five of these seven held their peg. Falcon's USDf is the mild exception and it is worth stating precisely rather than dramatically: its supply barely moved over the month while its price sits about half a percent under a dollar, which is a different and much smaller thing than a depeg. USR is the real exception, and it is the reason it sits apart in the reading column: its supply left because the token broke in an off-chain mint exploit we walked separately, not because demand cooled. Resolv has now fallen another 52 percent in thirty days and sits 99.1 percent below its peak.

Two protocols in the field finished the job this month. Elixir, which peaked above $342M in July 2024, now shows $13.00 of total value, down 100.000 percent from that peak, and its own documentation states that deUSD holds no value and has been sunset. Umoja's yBTC went from about $60,000 seven days ago to $0.00. Neither is a depeg either. Both are the end state of the same demand story, reached faster.

Why supply leaves while the price holds

A delta-neutral synthetic dollar pays its yield from two places: the funding rate earned on the short-perpetual hedge, and whatever the collateral itself earns. Both compress in the same regime. As perpetual funding normalized off its highs, the headline yields that pulled capital in came down with it; Ethena's staked sUSDe, the benchmark for the whole category, now sits in the high threes on Ethena's own dashboard, roughly the level of short-term Treasuries and a long way from the double digits of a hotter market. At the same time, many of these dollars grew on points and airdrop incentives, and those programs run out. When the yield compresses and the points stop, the capital that came for the yield and the points does the rational thing and leaves, at par, through a redemption door that is working exactly as it should. That is the entire mechanism. It is the same shape we walked in the MegaUSD supply collapse and in Neutrl's NUSD: a redemption run at par, not a depeg. Peg failure is the tail risk everyone watches for. Yield compression is the ordinary stress the category is living through right now, and it is quieter precisely because nothing breaks.

You do not have to take the compression on faith

The claim that yields have compressed is usually made from marketing pages, which is the worst possible source for it. So we read it from the vaults instead. A share price is the one figure a synthetic dollar cannot spin: it is a public read from the protocol's own contract, it settles that protocol's own redemptions, and anyone can re-read it at a past block. Our Honesty Index does that hourly, for every protocol on it, and publishes what each one advertises beside what its own vault actually did.

Read that board today and the compression is not a thesis, it is a column. The live protocols cluster between roughly two and nine percent realized over thirty days, with the top of that range belonging to the one dollar in the table above that grew. More striking are the vaults where accrual simply stopped: Level's slvlUSD share price has not moved since October 31, 2025, and Level's own documentation now carries a redeem-only notice; Stables Labs' sUSDX has not moved since November 7, 2025. Both were verified against two independent archive providers at seven, thirty, ninety and one hundred eighty days. Neither of those is an accusation. A protocol that announces a wind-down, stops accruing, and tells holders to leave is behaving correctly, and a frozen share price is the honest record of it.

The board is deliberately narrow about what it will assert. It scores one axis, yield, and it refuses to difference a seven day advertised figure against a thirty day realized one, because that manufactures a discrepancy out of a unit mismatch. Every advertised number on it comes from the issuer's own surface, quoted and dated, never from an aggregator, for a reason that has already bitten once: aggregators list figures these protocols never published, and printing one as theirs would be the exact error the page exists to criticise.

The biggest name is diversifying out of the pure basis trade

The clearest sign that the category is repricing yield is what its largest member is doing with its own backing. Ethena spent this year diversifying USDe away from the crypto basis trade that defined it. By its April 2026 reserve overhaul, perpetual-futures positions, once the mainstay of USDe's backing, were down to roughly eleven percent of reserves, with the rest in stablecoin reserves, DeFi and institutional lending, and real-world assets like tokenized Treasuries and short-duration credit. Ethena is still the largest synthetic dollar in the market by a wide margin, and it is extending basis trades into new markets like equity and commodity perpetuals, so this is a diversification of its backing rather than a withdrawal from the trade. But the pure crypto-funding basis trade, the mechanism this entire category was built on, is now a minority of the biggest issuer's book, and its staked yield has compressed to about the level of Treasuries. The pure basis lane is thinner now, and its largest occupant stands mostly outside it.

What this is not

This is not a prediction that these dollars are about to depeg, and it is not a claim that any of them is under-backed. The peg holding through a supply contraction is evidence that the redemption path works, which is the good outcome, not the bad one. Ethena in particular is the largest synthetic dollar in the market with a real user base, and a shrink from a frothy peak is not a failure. Nor is the contraction universal: two dollars in the table above grew about ten percent over the same thirty days, and any story that needs the whole category to be failing is already wrong. This is a demand-and-yield observation, not a backing accusation. The reason to write it down is that "the category is shrinking" is about to become conventional wisdom, and when it does, the lazy version of it will blur the two kinds of shrink together. They are not the same, and the difference is the whole thing.

The part that applies to us

We are the smallest dollar in this analysis by orders of magnitude, around eleven hundred dollars of kUSD outstanding and pre-audit, and essentially all of it is ours rather than third-party demand, so none of this is a claim to be winning a contraction we are barely inside. But the physics are the same for us, and pretending otherwise would be the exact dishonesty this piece is about. kUSD's yield is funding-rate dependent. In the same regime that is compressing everyone else's, ours compresses too. That is precisely why the rate we publish is a model of the book we actually run rather than a promise. It is recomputed from live funding on every read, and because the hedge is sized one for one against on-chain spot, the multiplier on the carry sits below one, so the number falls as funding falls instead of being held up by a leverage constant. It reads in the low single digits today.

There used to be a second, larger figure published beside it, a modeled target at scale in the low teens, carrying its own label. On July 28, 2026 we withdrew it, and it is not coming back. It was reachable only if the spot leg were itself levered to three times notional through a borrowing facility we do not operate, the model charged no borrow cost against that leverage, and no holder could ever have been paid it. Our published through-cycle band, 8 to 9.4 percent, is what this design could sustain with leverage and a funded insurance buffer behind it, which puts it above today's deployment rather than under it. The method is written out at yield methodology, next to a live delta gauge and a realized-distribution ledger that only recently began rather than a projected number. We would rather show a smaller honest range that survives a compression than a large one that does not. The compression is not a surprise we are managing around. It is a regime we priced into how we describe ourselves.

And because the board above is ours, the fair thing is to say where we land on it. Our own advertised figure is a model and our own realized figure is close to nothing, which makes the distance between them one of the widest on a page we built and published ourselves. That is not modesty. It is the reason the other rows are worth reading. There is one more thing worth saying, and we will say it carefully because we have not earned a victory lap at eleven hundred dollars of size. As the largest name diversifies out of the pure crypto basis trade, the transparent, pure-play version of that trade is getting rarer, and at Genesis scale and pre-audit that narrow lane is the one kUSD is in. That is an observation about where the category is heading, not a claim to be ahead in it.

How to read a shrinking category

If you allocate to synthetic dollars, the useful response to this month is not to flee the category. It is to change what you judge them by. Ask whether a dollar's yield survives a full funding cycle or only exists at the top of one, because a through-cycle number tells you far more than a headline. Ask whether its supply is sticky or rented, because capital that arrived for points leaves for points, and a supply chart tells you which one you are looking at. Ask what its vault share price actually did, not what its front page says, because those are different numbers and only one of them settles your redemption. And ask whether you can recompute the backing yourself, because the one kind of shrink that should scare you, the Resolv kind, is invisible on a marketing page and obvious on-chain. You can run the yield half of that on the Honesty Index, the reserve half on anything with /verify-anything, see the whole field ranked in the scorecard, and read the ones that did not make it in the graveyard.

Figures are as of the publication date and nothing here is investment advice. Stablecoin prices and circulating supplies for USDe, USDf, USD0, NUSD, USR, avUSD and trUSD are per DefiLlama, read on the publication date, and the one-month changes are computed against the same series. On the read behind this piece, USDe circulating was about $3.845B, within roughly 0.04% of a dollar and down about 13.8% over the trailing month from about $4.462B; USDf about $1.255B at about $0.9947 and flat over the month; USD0 about $553M near par; NUSD about $58.6M near par and down about 27.5% from about $80.8M; USR about $968K at about $0.1585; avUSD about $116.8M and up about 9.9% from about $106.4M; trUSD about $54.1M and up about 9.5%. The category total is the sum of all forty protocols DefiLlama classifies as Basis Trading, recomputed from each protocol's own TVL series rather than read off a category page: about $6.936B today against about $7.652B thirty days ago, a fall of about 9.4% or about $716M. Elixir's $13.00 and its $342.5M July 2024 peak, Umoja yBTC's move to $0.00, and Resolv's 30-day and peak-to-now declines are from the same per-protocol series. The sUSDe benchmark yield is Ethena's own published staking rate at app.ethena.fi, read on the publication date, and funding-rate normalization is observable on public perpetual venues. That Ethena reduced the crypto perpetual-futures basis trade to about eleven percent of USDe's backing in its April 2026 reserve overhaul, diversifying into stablecoin reserves, lending and real-world assets, is per The Defiant and Ethena's own disclosures. Every realized yield figure, including the frozen share prices for Level's slvlUSD and Stables Labs' sUSDX and the dates they last moved, is computed from ERC-4626 convertToAssets read at two block heights and is published with those block numbers on the Honesty Index, where anyone can re-read them. The Resolv USR exploit figures are from its own retrospective. Kerne's own yield is a live model, not a promise: the through-cycle band, the method, and the realized-distribution ledger are at /docs/yield-methodology, and the live risk surface is at /api/risk-status. A /verify pass proves an attestation is authentic and fresh; it is not an audit and not a solvency opinion.