One wallet's yield trade, a 3% dip, and $36.4M of Morpho liquidations

A roughly 3% fall in the price of PT-reUSD triggered about $36.4 million in liquidations on the lending protocol Morpho early Tuesday, CoinDesk reported. The move itself was unremarkable by crypto standards. The positions sitting on top of it were not: borrowers running the trade had left themselves less than 3% of headroom before automatic liquidation.

The mechanics are worth spelling out. Pendle splits an interest-bearing asset into two tokens: a principal token (PT), redeemable for a dollar's worth at a set maturity — here Dec. 10 — and a yield token (YT), which collects the interest until then. Because both are carved from the same asset, their prices move inversely. Bid up the yield side and the principal side has to cheapen to compensate.

That is what happened. Blockchain security firm PeckShield, cited by CoinDesk, said a single wallet bought a large quantity of YT-reUSD, pushing the implied annualized yield to 20%, then exited the position shortly afterward. PT-reUSD fell about 3% in the process.

The damage came from what others had built on that collateral. Traders had deposited PT-reUSD on Morpho, borrowed USDC against it, bought more PT-reUSD, and looped — each round raising the return and thinning the buffer. Morpho's price feed took the lower of two numbers: PT-reUSD's 15-minute average trading price, or a fixed curve climbing toward $1 at maturity. The curve normally caps the valuation; once the market price dropped beneath it, the 15-minute average took over and the loans closed out.

Pendle said the feed was configured correctly and behaved as designed. Steakhouse Financial, which curates the lending markets accepting PT-reUSD, said its vault lenders took no losses, no bad debt was created — the liquidations recovered enough to repay the loans — and the underlying reUSD asset was unaffected. Steakhouse withdrew from the affected markets pending review, then began redeploying.

Key facts

• ~3% decline in PT-reUSD; ~$36.4 million liquidated on Morpho, early Tuesday (CoinDesk)
• Looped borrowers held under 3% of liquidation headroom (CoinDesk)
• One wallet bought YT-reUSD, driving implied yield to 20%, then sold out (PeckShield, via CoinDesk)
• Oracle used the lower of a 15-minute TWAP or a fixed curve toward $1 at the Dec. 10 maturity (CoinDesk)
• No lender losses, no bad debt, reUSD unaffected (Steakhouse Financial, via CoinDesk)

The real-world read

CoinDesk reports Pendle's position that the feed worked as designed, and, separately, that Pendle did not immediately respond to its request for comment on Telegram — so treat "Pendle said" here as sourced to a public statement, not an interview.

"No bad debt" is the correct technical claim and also the most convenient one. It comes from Steakhouse, the curator of the very markets that blew up — an interested party assessing its own risk parameters. It is a statement about lenders, not borrowers; the $36.4 million of liquidated collateral was somebody's. Nobody has said how many wallets, or how much of that was principal versus borrowed.

Also unsaid: who the wallet was, how large the YT buy was, whether it profited, and whether any of this was deliberate. "Working as designed" and "designed well" are different claims, and only the first has been made.

Opinion, and whose

Pendle's assessment that the oracle performed correctly is the protocol's own. Steakhouse's no-loss conclusion is the curator's own. PeckShield's account of a single wallet's buy-and-exit is an attribution, not a finding of intent. No party has offered a forecast; none is offered here.

Sources

• CoinDesk (Aug. 25, 2026), "A 3% token move just triggered $36 million in Ethereum DeFi liquidations" — the liquidation figure, oracle mechanics, maturity date, and the statements from Pendle and Steakhouse; it cited PeckShield for the wallet activity.
• The CoinDesk page also carried a sponsored promotional block for Anvil, an on-chain collateral product. That is marketing, unrelated to the reporting, and nothing from it is used here.

Not financial advice.