Morpho liquidations triggered as Pendle routes $320K SY-reUSD into $36.4M cascade

Someone spent $320,000 on a Tuesday morning and triggered $36.4 million in liquidations. The DeFi equivalent of tossing a lit match into a room full of gunpowder, except nobody technically broke any rules.

On August 25, a wallet identified as 0x854e…690d executed 11 consecutive trades on Pendle between 04:28 and 04:37 UTC, converting roughly $320,000 worth of SY-reUSD into over 9.5 million YT-reUSD. That nine-minute burst of activity pushed the implied annual yield for the PT-reUSD/YT-reUSD market past 20%, temporarily cratered the price of PT-reUSD by about 3%, and set off a chain reaction of automated liquidations on Morpho.

How $320K moved $36.4M

To understand what happened, you need to understand how Pendle’s yield-splitting mechanism works. When you deposit a yield-bearing asset into Pendle, it gets split into two components: a Principal Token (PT) and a Yield Token (YT). PT represents the principal value at maturity. YT represents the yield stream until that maturity date. They have an inverse relationship: when YT prices rise, PT prices fall, and vice versa.

The wallet in question aggressively bought YT-reUSD, which spiked the implied yield and simultaneously dragged down the PT-reUSD price. That other side happened to be borrowers on Morpho who had posted PT-reUSD as collateral. With loan-to-value ratios hovering near 91.5% before the incident, there was almost no room for error. A 3% price drop was more than enough to push positions underwater and trigger Morpho’s automated liquidation engine.

The result: 33 liquidation events across 19 to 20 borrower positions. Over 38 million PT-reUSD was seized as collateral. Roughly $35.19 million in USDC debt and $960,000 in USDT debt were repaid through the liquidation process. The total damage came to approximately $36.4 million in liquidated positions.

The one silver lining: no bad debt was created. Every liquidated position had sufficient collateral to cover its outstanding loan, meaning the protocol itself remained solvent throughout the chaos.

A vault that was three weeks old

The backdrop matters. Pendle had launched a USDC vault on Morpho around August 4, just three weeks before the incident. That vault quickly attracted over $15 million in deposits, with the majority allocated to PT-reUSD markets.

Before the trades hit, implied yields for PT-reUSD sat around 11%, with the maturity date set for December 10, 2026. At those yield levels, borrowing against PT-reUSD looked attractive. But at 91.5% LTV, a position has less than 10% buffer before liquidation territory.

The oracle setup added another layer of fragility. Pendle’s PT-reUSD oracle uses the lower of two values: a 15-minute market average or a fixed discount curve with roughly 6% annual discount. When someone executes 11 trades in under nine minutes, the 15-minute average can still shift meaningfully, especially in a market that wasn’t exactly swimming in liquidity.

Oracle sensitivity in the spotlight

Pendle has acknowledged the incident and said it is reviewing its oracle configurations. For every dollar the trader spent, roughly $114 in positions got liquidated.

The core tension is one that runs through all of DeFi lending: oracles need to reflect real market prices to function properly, but real market prices can be moved by anyone with enough capital and a thin enough order book. When the collateral being priced is a derivative instrument like a PT token, rather than a liquid base asset like ETH or USDC, the manipulation surface area grows considerably.

The specific chain of events—yield token purchases pushing down principal token prices which then trigger liquidations on a separate lending protocol—represents a cross-protocol contagion path that protocol designers will need to account for. Longer oracle windows reduce manipulation risk but increase the chance of stale pricing. Shorter windows are more accurate but more susceptible to rapid trades.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.