Aave PT rollover: when borrowing costs beat the yield

What is being proposed

The plan would move collateral from the PT maturing Oct. 29 into a replacement maturing Feb. 25, 2027. A flash loan would temporarily repay the debt, swap the old PT for the new one, borrow again and close the flash loan in one step. OKX would build that integration separately.

A principal token, or PT, is a claim on an asset at maturity bought at a discount. The discount is the yield. Loopers borrow stablecoins against the PT, buy more PT, and repeat. The trade works only while the PT yield exceeds the borrowing rate. That spread is the carry.

How the carry flipped

In August the trade looked comfortable. On Aug. 18 figures, LlamaRisk saw the October PT at 3.19%, while borrowing cost 0.45% in GHO, 1.37% in USDT0 and 2.69% in USDG. It described that as positive carry.

The February token is a different picture. Risk firm LlamaRisk found in an Oct. 8 assessment that its on-chain yield by itself undercut the cost of borrowing USDT0, USDC or USDG; only GHO was cheaper. The Oct. 7 snapshot showed the February PT at 2.79%, against 4.48% in USDT0, 4.10% in USDC and 3.67% in USDG, with GHO at 0.93%. An Oct. 10 refresh put the PT implied yield at about 2.98%, while variable borrow rates on Aave X Layer read USDT0 4.12%, USDC 5.87% and USDG 3.75%.

There is a further catch. USDG is not on the borrowable list LlamaRisk proposed for the new E-Mode 9 category, which names USDT0, GHO and USDC. Its rate is a comparison, not an available route.

What leverage does to a negative spread

CryptoSlate calculated that the 2.98% annual yield equals roughly 1.11% gross over about 137.1 days to Feb. 25. It then ran an illustration with fixed rates and no rewards.

At 2x exposure, return on equity stayed slightly positive: +0.69% borrowing USDT0, +0.05% with USDC and +0.83% with USDG. At 5x exposure it turned negative: -0.56% with USDT0, -3.11% with USDC and -0.02% with USDG. Financing cost grew faster than the PT gain.

This is the core mechanism. Each extra turn of leverage adds PT gain on a larger position, but it also adds financing on the borrowed amount. Once financing outweighs the gain, more leverage widens the loss on equity, before any price risk is counted. Execution costs make it thinner still: CryptoSlate listed 0.05% taker fees in the February pool, plus slippage, network fees and any flash-loan charge.

Liquidation limits and who is exposed

LlamaRisk suggested E-Mode 9 with a maximum loan-to-value of 91.48%, a liquidation threshold of 93.48% and a supply cap of 35 million PT. The current E-Mode 7 threshold is 94.66%. A lower threshold leaves less room before liquidation for the same debt. On Aave, liquidation becomes possible once a position's health factor drops below one.

The existing cohort is tightly levered. A Sept. 9 borrower review counted 55.1 million USDT0 of debt, 98.7% of the cohort total, and the top suppliers sat at a median health factor of just 1.03. October PT collateral supplied was 66.03 million units per LlamaRisk on Oct. 8 and 59.79 million on the Aave display on Oct. 10, when total X Layer borrowing was $61.12 million. The February pool held only about $1.42 million of reported liquidity, which limits how much can migrate at acceptable slippage.

LlamaRisk credited loopers' profits to external OKX Exchange Earn incentives and said their end would likely push users to unwind. A separate OKX Wallet bonus, capped at 2% APY for the October token, ends Oct. 28 at 07:00 UTC, and payouts may fall below the announced 490,000 USDT pool. The risk types here are smart-contract, liquidity and rate risk, stacked on each other.

• Does my carry still hold at today's borrow rate, not the rate when I opened?
• How much of my return depends on rewards that have an end date?
• Under the proposed threshold, how far is my health factor from one?
• Can I exit through a pool with about $1.42 million of liquidity without large slippage?

What to watch next

The rollover is still a proposal, and OKX would build its integration separately. Watch whether E-Mode 9 is adopted with the suggested limits, how variable borrow rates move against the PT yield, and what happens to looped positions as the Oct. 28 bonus deadline and the Oct. 29 maturity arrive.