Sometimes the most unsettling incidents involving protocols like Pendle aren’t the ones where something breaks.
They’re the ones where everything works exactly as designed, and people still lose tens of millions of dollars. That’s precisely the situation Pendle and Steakhouse Finance found themselves explaining this week, after a sharp, brief price move on a single collateral asset triggered $36.1 million in liquidations, and at least one researcher believes someone walked away having engineered the whole thing.
What Actually Happened In The Early Morning Hours
The event centered on the PT-reUSD/USDC market, a lending market deployed by Steakhouse Finance on Morpho using Pendle’s PT-reUSD asset as collateral. Pendle’s own team confirmed the incident directly, stating plainly that its oracle for this specific market was set up correctly and functioned exactly as intended, explicitly ruling out a misconfiguration despite the unfortunate outcome.
According to Steakhouse’s own account, at around 04:30 UTC, price action around the PT-reUSD asset, tied to a December 10 maturity, caused a number of large, highly leveraged positions to liquidate. Steakhouse was direct about the scale involved too, noting this kind of outcome becomes increasingly likely as trade size grows.
How Pendle’s Oracle System Actually Works
I think understanding the mechanics here is essential to judging whether this was a genuine failure or simply a design working under pressure. Pendle’s oracle setup for its PT tokens varies by asset and market, and in PT-reUSD’s specific case, Steakhouse configured the oracle to reference whichever of two prices was lower: the PT’s own market price on Pendle, taken as a 15-minute average, or a fixed curve that gradually rises to $1 at maturity along a 6% annual discount rate.
At the peak of the incident, the PT price dropped by roughly 3%, and that 15-minute market average took over as the active oracle reference instead of the fixed curve. That shift lowered the collateral value of looped PT positions that had been borrowing USDC against it on Morpho. Positions already sitting with health factors below 1.03, meaning they had less than a 3% buffer against any price movement at all, dropped below the liquidation threshold almost immediately once that repricing happened.
Pendle And Steakhouse Say No Lasting Damage Was Done
Both teams were emphatic that the system contained the damage as intended. Pendle confirmed its Ecosystem Vault on Morpho was not affected, that liquidations processed normally, and that no bad debt was incurred anywhere in the process. Steakhouse echoed that directly in its own update, stating no lenders in its vaults were affected, no bad debt occurred, and the underlying reUSD asset itself remained entirely unaffected throughout.
Steakhouse also described taking a precautionary step once the volatility began, saying its systems withdrew liquidity from all affected markets while the situation was being evaluated, and that liquidity was being restored back into those markets once the immediate risk had passed. Pendle framed the underlying oracle design choice, combining a 15-minute time-weighted average price with a linear discount curve, as generally good practice specifically because it’s meant to protect lenders against bad debt in the event the underlying asset becomes impaired. In other words, both teams are arguing the mechanism that triggered these liquidations is the same mechanism that kept the losses contained to leveraged borrowers rather than spreading into lender funds.
A Very Different Read From Herd’s Co-Founder
Here’s where the story gets genuinely more complicated. Herd co-founder Andrew Hong offered a considerably sharper characterization of the same event, describing it not as an unfortunate but ordinary market move, but as oracle manipulation. According to his analysis, 12 borrowers holding a combined $36.1 million in positions were liquidated as a result of what he specifically called a 2.6% PT-reUSD oracle manipulation, with the party responsible capturing 96% of the liquidated collateral, netting approximately $1.3 million in profit.
Hong added a technically important detail too: the TWAP-based oracle configuration was specifically chosen for reUSD markets deployed on June 15 onward, whereas markets deployed prior to that date used the linear discount oracle exclusively. That distinction matters, because it suggests the specific oracle setup that allowed a 15-minute average to briefly override the discount curve wasn’t universal across every PT-reUSD market, it was a configuration choice made for this particular batch of markets.
Where Both Accounts Actually Agree
Despite the sharply different framing, both Steakhouse and Herd’s own findings agree on the parts that matter most for anyone with money in these markets right now. Hong confirmed that supplying vaults were entirely unaffected, and said he didn’t find any vaults among the liquidated borrowers, meaning no contagion into the broader lending pool should be expected. That lines up directly with what Pendle and Steakhouse both stated publicly: lender positions were protected, and no bad debt resulted from the event.
Where the accounts diverge is entirely on intent and characterization. Pendle and Steakhouse describe a market functioning exactly as its risk parameters were designed to function, with an unfortunate but expected outcome for over-leveraged positions during a genuine price move. Hong’s framing suggests something more deliberate, a price movement engineered specifically to push thin-margin positions below their liquidation threshold, with the profit flowing disproportionately to whoever triggered it.
What This Means For Anyone Using Leveraged PT Positions
I think the honest takeaway here is that both things can be true simultaneously, and that’s precisely what makes this incident worth taking seriously rather than dismissing either account outright. The oracle mechanism itself did exactly what it was built to do, protecting lenders and preventing bad debt, and Pendle’s broader design philosophy of using the lower of two price references is a genuinely sound piece of risk engineering. But a mechanism working as designed doesn’t automatically mean nobody exploited the specific conditions that mechanism creates, and a 96% capture rate on liquidated collateral going to what appears to be a single party is a genuinely striking statistic that deserves more scrutiny than a routine liquidation event would.
For anyone running leveraged, looped positions against PT collateral on Morpho or similar protocols, I think the real lesson here isn’t about whether Pendle’s oracle design is sound, it largely appears to be. It’s about how thin a margin actually is once you’re borrowing against a position with a health factor sitting at 1.03 or below. A 2.6% to 3% price move was all it took to wipe out $36 million in positions in a matter of minutes, and whether that move was organic market volatility or something more deliberately triggered, the underlying vulnerability, razor-thin liquidation buffers on highly leveraged looped positions, was exactly the same either way.
Disclosure: This is not trading or investment advice. Always do your research before buying any cryptocurrency or investing in any services. Follow us on X @nulltxnews
Be the first to comment