Yield Magnetism and System Resilience: How to Rationally View the YT Arbitrage of Ethena × Pendle

How Arbitrage Works: From "Yield Side" to "System Side"

Let's first look at this arbitrage path: deposit eUSDe or sUSDe (sUSDe is the staked eUSDe, yielding native returns) in Aave, borrow stablecoins, and then buy YT/PT in Pendle. YT corresponds to future yields, while PT can always be bought at a discount because it has stripped away the yield, held until maturity to redeem at 1:1, earning the price difference. Of course, the real "big gain" comes from external incentives like Ethena points.

The PT obtained this way, since it can serve as collateral in Aave, becomes the perfect starting point for circular loans: "Collateralize PT → Borrow stablecoins → Buy PT/YT → Re-collateralize." This is done to leverage relatively certain yields to pursue high-elasticity returns like Ethena points.

How has this capital chain rewritten the lending market?

Aave's Exposure and Second-order Effects: Assets supported by USDe have gradually become mainstream collateral on Aave, with its share rising to about 43.5%, directly boosting the utilization rates of mainstream stablecoins USDT/USDC.
Crowding on the Borrowing Side: After introducing USDe eMode for PT collateralization, the borrowing scale of USDe reached approximately $370 million, of which about $220 million (≈60%) serves leveraged PT strategies, with utilization soaring from about 50% to around 80%.
Concentration and Re-collateralization: The supply of USDe on Aave is highly concentrated, with the top two entities accounting for over 61%. This concentration, combined with circular leverage, amplifies yields and exacerbates the system's fragility.

The rule here is simple: the more enticing the yield, the more crowded the cycle, and the more sensitive the entire system becomes. Any slight fluctuation in price, interest rates, or liquidity will be ruthlessly amplified by this leverage chain.

Note: The core on-chain data cited in this article is primarily based on the report released by Chaos Labs on July 17, 2025, and related market observations. Due to the dynamic nature of on-chain data, readers are advised to check the latest situation through relevant data analysis platforms.

Why "Exiting" Becomes Difficult: Pendle's Structural Constraints

So, how to exit? When reducing leverage or closing positions in the aforementioned circular positions, there are mainly two paths:

Market Exit: Sell PT/YT before maturity to exchange for stablecoins to repay and release collateral.
Hold Until Maturity Exit: Hold PT until maturity, redeeming the underlying asset at 1:1 to repay. This path is slower but more stable during market fluctuations.

Why does exiting become difficult? The challenges mainly stem from two structural constraints of Pendle:

Fixed Term: PT cannot be directly redeemed before maturity and can only be sold in the secondary market. To "quickly reduce leverage," one must watch the secondary market and endure the dual challenges of depth and price fluctuations.
AMM's "Implied Yield Range": Pendle's AMM operates most efficiently within a preset implied yield range. If market sentiment changes and the yield pricing exceeds this range, the AMM may "deactivate," forcing trades to occur on thinner order books, sharply increasing slippage and liquidation risks. To prevent risk spillover, protocols like Aave deploy PT risk oracles: when the PT price drops to a certain floor price, the market is directly frozen. This can prevent bad debts but also means that you may find it difficult to sell PT in the short term, having to wait for the market to recover or hold until maturity.

Thus, exiting is usually not difficult when the market is stable, but when the market begins to reprice and liquidity becomes crowded, exiting becomes a major friction point that requires advance preparation.

Aave's "Brakes and Buffers": Making De-leveraging Orderly and Controllable

In the face of such structural frictions, how do lending protocols (like Aave) conduct risk control? They have built-in a set of "brakes and buffers" mechanisms:

Freezing and Floor Price Mechanism: If the PT price touches and maintains the oracle's floor price, the relevant market can be frozen until maturity; after maturity, PT naturally decomposes into the underlying asset, allowing for safe liquidation/release, aiming to avoid liquidity misalignment spillover caused by fixed-term structures.
Internalized Liquidation: In extreme cases, the liquidation reward is set to 0, forming a buffer before segmenting the disposal of collateral: USDe is sold in the secondary market after liquidity recovery, while PT is held until maturity to avoid being passively sold on a thin order book in the secondary market, thereby amplifying slippage.
Whitelist Redemption: If the lending protocol obtains an Ethena whitelist, it can bypass the secondary market and directly redeem the underlying stablecoins with USDe, reducing impact and enhancing recovery.
Boundary of Supporting Tools: When USDe liquidity is temporarily tight, Debt Swap can convert USDe-denominated debt into USDT/USDC; however, due to E-mode configuration constraints, migration has thresholds and steps, requiring more sufficient collateral.

Ethena's "Adaptive Base": Supporting Structure and Custody Isolation

While lending protocols have "brakes," the asset support side needs Ethena's "automatic transmission" to absorb shocks.

On the supporting structure and funding rate status: When funding rates decline or turn negative, Ethena reduces hedge exposure and increases stablecoin support; in mid-May 2024, the stablecoin ratio once reached ~76.3%, then fell back to ~50%, still high compared to previous years, allowing for proactive pressure reduction during negative funding rate periods.
Further, from the perspective of buffering capacity: In extreme LST confiscation scenarios, the overall net impact on USDe support is estimated at about 0.304%; a reserve of $60 million is sufficient to absorb such shocks (only about 27% of it), thus the substantive impact on anchoring and repayment is controllable.
The custody and isolation of assets is a key link: Ethena's assets are not directly stored in exchanges but are settled and isolated through third-party custodians (like Copper, Ceffu) for over-the-counter settlement and asset isolation. This means that even if the exchange itself encounters operational or repayment issues, these assets serving as collateral are independent and protected in ownership. Under this isolation structure, efficient emergency processes can be realized: if an exchange is interrupted, custodians can invalidate unclosed positions after missing a certain number of settlement rounds, releasing collateral to help Ethena quickly migrate hedge positions to other exchanges, significantly shortening the risk exposure window.

When misalignment mainly comes from "implied yield repricing" rather than damage to USDe support, under the protection of oracle freezing and layered disposal, the risk of bad debts is controllable; the real focus should be on preventing tail events that damage the support side.

What You Should Pay Attention To: 6 Risk Signals

Having discussed the theory, what specific indicators should we look at? The following six signals are highly correlated with the interaction of Aave × Pendle × Ethena and can serve as a daily dashboard for monitoring.

USDe Borrowing and Utilization Rate: Continuously track the total borrowing amount of USDe, the proportion of leveraged PT strategies, and the utilization rate curve. When the utilization rate remains above ~80%, the system's sensitivity significantly increases (the reporting period rose from ~50% to ~80%).
Aave Exposure and Stablecoin Second-order Effects: Pay attention to the proportion of USDe-supported assets in Aave's total collateral (e.g., ~43.5%) and the transmission effect on the utilization rates of core stablecoins like USDT/USDC.
Concentration and Re-collateralization: Monitor the deposit proportion of top addresses; when the concentration of top addresses (e.g., the top two combined) exceeds 50-60%, be wary of potential liquidity shocks caused by their synchronized operations (reporting period peak >61%).
Proximity to Implied Yield Range: Check whether the implied yield of the target PT/YT pool is approaching the boundary of the AMM preset range; proximity or exceeding the range indicates decreased matching efficiency and increased exit friction.
PT Risk Oracle Status: Pay attention to the distance between the market price of PT and the minimum price threshold of Aave's risk oracle; nearing the threshold is a strong signal that the leverage chain needs to "orderly decelerate."
Ethena Support Status: Regularly check the reserve composition published by Ethena. Changes in the stablecoin ratio (e.g., falling from ~76.3% to ~50%) reflect its adaptive strategy to funding rates and system buffering capacity.

Furthermore, you can set trigger thresholds for each signal and plan response actions in advance (e.g., utilization rate ≥80% → reduce circular multiples).

From Observation to Boundaries: Risk and Liquidity Management

These signals ultimately serve risk control. We can solidify them into four clear "boundaries" and operate around the closed loop of "risk limits → trigger thresholds → disposal actions."

Boundary 1: Circular Multiples

Circular leverage amplifies sensitivity to price, interest rates, and liquidity while enhancing returns (especially when combined with external incentives); the higher the multiple, the smaller the exit margin.

Limits: Set a maximum circular multiple and minimum collateral redundancy (e.g., LTV/Health Factor lower limit).

Triggers: Utilization rate ≥ 80% / Stablecoin borrowing rate rises rapidly / Proximity to range increases.

Actions: Reduce multiples, supplement collateral, pause new cycles; if necessary, switch to "hold until maturity."

Boundary 2: Term Constraints (PT)

PT cannot be redeemed before maturity; "hold until maturity" should be viewed as the regular path rather than a temporary measure.

Limits: Set a scale limit for positions relying on "selling before maturity."

Triggers: Implied yield exceeds the range / Market depth plummets / Oracle floor price approaches.

Actions: Increase cash and collateral ratios, adjust exit priorities; if necessary, set a "only reduce, not increase" freeze period.

Boundary 3: Oracle Status

Price nearing the minimum price threshold or triggering a freeze indicates that the link is entering an orderly deceleration phase for de-leveraging.

Limits: Minimum price difference (buffer) with the oracle floor price and shortest observation window.

Triggers: Price difference ≤ preset threshold / Freeze signal triggered.

Actions: Segment liquidation, raise liquidation alerts, execute Debt Swap / de-leveraging SOP, and increase data polling frequency.

Boundary 4: Tool Friction

Debt Swap, eMode migration, etc., are effective during tight periods but come with thresholds, waiting, additional collateral, and slippage frictions.

Limits: Available quota/time window for tools and maximum tolerable slippage and costs.

Triggers: Borrowing rate or waiting time exceeds threshold / Trading depth falls below the lower limit.

Actions: Reserve funding redundancy, switch to alternative channels (gradual liquidation/hold until maturity/whitelist redemption), and pause strategy expansion.

Conclusion and Future Directions

In summary, the arbitrage between Ethena and Pendle has woven Aave, Pendle, and Ethena into a transmission chain from "yield magnet" to "system resilience." The circular push from the capital side has heightened sensitivity, while the structural constraints on the market side have raised exit thresholds, and protocols provide buffers through their respective risk control designs.

In the DeFi space, the advancement of analytical capabilities lies in how we view and use data. We are accustomed to using data analysis tools like Dune or DeFiLlama to review the "past," such as tracking changes in positions of top addresses or trends in protocol utilization rates. This is important as it helps us identify system vulnerabilities like high leverage and concentration. However, its limitations are also evident: historical data presents a "static snapshot" of risks but cannot tell us how these static risks will evolve into dynamic system collapses when market storms arrive.

To see these hidden tail risks clearly and deduce their transmission paths, we need to introduce forward-looking "stress tests" — this is precisely the role of simulation models. They allow us to parameterize all the risk signals mentioned in this article (utilization rates, concentration, prices, etc.) and place them into a digital sandbox (a joint model composed of the core mechanisms of Aave, Pendle, and Ethena), repeatedly asking "What if…?"

If ETH price drops by 30% while funding rates turn negative, how long can my position hold?
To exit safely, how much slippage do I need to endure?
What should be the minimum safe collateral?

The answers to these questions cannot be directly found from historical data but can be anticipated through simulation modeling, ultimately helping you form a truly reliable execution manual. If you want to practice, you can choose the industry-standard framework cadCAD based on Python or try the next-generation platform HoloBit based on cutting-edge Generative Agent-Based Modeling (GABM) technology, which offers powerful visualization and no-code capabilities.