The looping trap

• One risk profile for the whole vault. Looping stacks leverage on the same yield source – every depositor ends up holding the loop, whether they wanted that risk or not.
• Yield comes from leverage, not underwriting. Loss is socialized: at typical 5–10× loop factors, a 5% hit on the underlying becomes a 25–50% hit on every depositor equally.
• Reflexive in stress. Collateral drops or borrow rates spike → loops unwind → forced selling and exit-queue stress → collateral drops further. The wstETH:ETH unwind on Aave (Jul 2025) ran this script in days – staking exit queues hit 9–18 days while spot stETH depegged.
• Phantom TVL hides insolvency. Recursive collateral inflates headline TVL beyond the underwriting that backs it. Stream Finance's xUSD/deUSD loops printed ~3.25× phantom TVL until a Balancer V2 rounding exploit (Nov 2025) detonated ~$285M of cross-protocol bad debt across Euler and Morpho.

What tranching changes

• Two contractual loss positions on the same underlying. Junior absorbs the first dollar; Senior is protected up to a defined cushion.
• Yield is priced against the position, not the leverage. Junior earns a risk premium for absorbing volatility; Senior earns a capped, protected coupon.
• Loss is ranked, not socialized. A 5% drawdown that hits a loop's depositors equally instead burns Junior's buffer while leaving Senior whole.
• Scale comes from underwriting, not leverage. The constraint moves from "how much can we lever" to "how much first-loss capital can we attract" – a fundamentally different question.

The structural flow

Before / after a 30% loss · $100 deposit

Junior absorbs every dollar of loss until exhausted. Senior is untouched up to the subordination boundary.

$100 across five states · illustrative

*Illustrative figures; actual srRoyUSDC ratios and APYs vary by vault and utilization. Royco's observation-period mechanic means transient mark-to-market dips don't trigger Junior loss recognition – only persistent declines count.

Mechanics

• Senior: "smart-contract-enforced downside protection." Junior absorbs losses from the very first dollar – no deductible.
• Junior: first-loss capital up to a defined coverage amount. Earns the underlying yield plus a risk premium paid by Senior in exchange for the protection.
• Yield split is dynamic, not fixed. A utilization curve (Aave-style) reweights yield toward whichever tranche is undersupplied – when Junior capital is scarce, more yield flows to Junior to attract first-loss capital, and vice versa.
• Observation period: a meaningful nuance – temporary mark-to-market drawdowns that revert don't trigger Junior loss recognition. Only persistent losses count, which prevents Junior wipeout from transient volatility.

Live vaults

A stablecoin issuer with a three-tier deposit hierarchy that mimics traditional fractional-reserve banking. Tranches are tied to stablecoin issuance, not wrapped vaults.

Loss waterfall

Explicit, coded: in a hack or bank-run, liUSD holders are wiped first, then siUSD, then iUSD. The longer the lock and higher the yield, the earlier in the firing line.

Underlying allocation

Aave, Pendle principal tokens, Ethena sUSDe, Fluid. A reserve buffer stays unallocated to support redemptions. Reportedly ~$1.60 of Ethena exposure per $1 deposited.

Mechanics

• Three positions, not two: Senior (Position 01) – pre-agreed fixed APY, paid first. Spectrum (Position 02) – user-selectable APY cap on a rate grid; multiple Spectrum tranches per pool, filled from lowest cap to highest. Junior (Position 03) – uncapped residual, first-loss from $1.
• Surplus Participation: when underlying performance exceeds Spectrum's caps, surplus goes to Junior first up to its hurdle, then is shared across Spectrum and Junior weighted by risk. This is the only protocol in the set with a mezz-style middle tier that earns an explicit kicker.
• Fixed maturities: each pool has a fixed maturity set at deployment. No early withdrawal from the pool itself, but every position is minted as a transferable ERC-20 so secondary markets can clear early exits.
• Loss waterfall: Junior absorbs from $1 → Spectrum (highest-cap first) → Senior last; enforced at maturity. Per Knox's explainer, optional pool-level safeguards (operator pause, early redemption) can exist if enabled at deployment.
• Underlying: curated yield from Morpho, Aave, Pendle, and other ERC-4626 vaults. Knox is a fixed-income chassis on top of established protocols, not its own yield source.

Three positions

How it works

• Junior tranche (Backers): individual underwriters who manually evaluate each Borrower Pool. Provide first-loss capital. Earn the headline borrower rate plus GFI rewards.
• Senior tranche (Senior Pool LPs): passive LPs deposit USDC into the Senior Pool and receive FIDU. The Senior Pool deploys across active Borrower Pools, levering Backer capital at ~4× by default – e.g., $200K of Backer capital pulls ~$800K of Senior Pool capital into the same pool.
• Repayment waterfall: every borrower payment pays Senior interest + principal first, then Junior. Backers miss repayment first on any shortfall.
• Yield split: Senior earns a lower net rate (leverage tax subsidizes Junior); Junior earns the headline rate plus underwriting rewards.
• Underlying: real-world private credit – historically loans to fintech lenders in emerging markets (Africa, LatAm, SE Asia).

Three documented defaults · The only real-world test

*Approximate figures based on public defaults documentation.