2026-09-19 · Stablecoins & yield

Avant via Royco — a second protective tranche on top of savUSD

Data as of 19 Sep 2026.

Bottom line. Royco slices savUSD — Avant's senior tranche — into two more pieces, and the second layer of protection is real: Royco's senior tranche absorbs Avant pool losses up to 34.85% versus 14.28% for plain savUSD, at a cost of ≈ 21bp a year. No protocol fee, no lockup. But it comes with two things plain savUSD doesn't have: the tranche logic has no published source code, and Royco's docs describe a different factory generation from the one actually running this market. Royco adds opposite things at once: more protection against economic loss inside Avant, and more code and key risk. The first only kicks in during a catastrophe; the second is live every day.

How much Avant pool loss is absorbed before the senior tranche starts losing. Royco's junior tranche covers the first 24% of losses at the savUSD level — another 20.6 percentage points measured against the pool. Royco market page and Avant dashboard, 19 Sep 2026.

What it is

Royco is a tranching protocol: it slices someone else's yield-bearing position into two pieces — the junior takes losses first, and the senior pays it for that with part of its yield. The market discussed here does this with savUSD, Avant's senior tranche. So the structure has two levels: inside Avant there's already a junior tranche protecting savUSD, and on top of it Royco adds another.

The "underlying" tab on the market page isn't a separate product or a collateral disclosure — it's an overview of how the base asset behaves. The only choice is senior or junior.

Double protection: the threshold rises 2.4x

Avant's waterfall, per its docs: reserve fund → junior tranche → senior tranche (savUSD). Royco inserts one more rung between Avant's junior tranche and the holder.

• The cost of protection has two readings almost 2x apart: 21bp from the market page itself (7.32% base → 7.12% senior, two independent samples, 19 Sep 2026) versus 38bp from the market curator's blog (17 Jul 2026, hedged with "approximately"). The 21bp figure is used here — it's live product data, not an estimate from the party paid to curate the market.
• The Royco senior share is accounted in dollars: on 19 Sep one share = 1.0371 avUSD (rate 0.862705 savUSD per share), i.e. +3.71% accrued since the market launched. No entry fee — the mint rate matches the current rate.
• The contract is named Royco Senior Tranche savUSD: a tranche, not a managed vault.
• Avant's exit trap: submitting a second withdrawal while one is pending resets the timer on the full amount. Exit in a single batch.

A five-of-anon multisig exists — but not on the tranche path

Yearn's public risk register (assessment 26 Mar 2026, overall 3.80 out of 5) found three multisigs in Royco: the owner is 3 of 5 anonymous addresses with no ENS, the treasury is the same five, there's no separation of powers, and it controls a function that sets reported asset size. No timelock.

Does this touch the senior tranche on savUSD? No — and on-chain reads settle it:

Fees check out too: only managed vaults take 10%; tranches pay zero.

🔴 But the market contract can't be read

Control: the same query against a known verified contract returns full source, so "unverified" is a fact, not a failed request.

• The docs describe a different deployment. The factory in Royco's docs and the factory running this market are different generations. The branch with money on it has a readable factory, and the unverified one is the one in the docs — which, if anything, reduces concern. But the consequence stands: that's why the token name (ROY-ST-) doesn't match what the UI promises (srRoy).
• The logic that computes shares and splits losses between senior and junior is still unreadable. A verified wrapper around unreadable logic doesn't fix that. The same flaw — instant, unverifiable upgrade rights — turned up twice, in two different layers, by two methods: a third-party audit at the vault level and on-chain reads at the market level. That's not one confirming the other: the objects are different.

Avant barely moved in three weeks

The reserve stays below 1% of the pool. savUSD share price read from the contract: 1.202141. Discrepancies left unaveraged: savUSD yield 7.32% on Royco's page vs 7.83% on Avant's dashboard (different windows); avUSD supply $127.2m in the app vs $127.04m on a third-party aggregator.

Wrapper depth

The chain becomes four links long: USDe → avUSD → savUSD → Royco tranche. A common DeFi rule is no more than three wrapper layers, hard no at four. Depending on whether USDe counts as a base asset or already as a wrapper over a basis trade, the structure sits exactly on the line or past it — the rule gives no clean answer here.

Who it suits, and when it breaks

• Suits: a savUSD holder whose main concern is economic loss in the Avant pool (≈ 75% USDe) and who's willing to trade 21bp a year and a doubled exit time for a buffer roughly 2.4x wider — accepting the risk of unreadable tranche code.
• Doesn't suit: anyone for whom code and key risk outweighs tail economic loss, or who needs to exit faster than 48 hours.
• What tips it toward Royco: verification of the market contract and the tranche implementation — then the main argument against disappears, and at 21bp for a wider buffer the structure looks rather attractive.
• The protection maths (14.28% → 34.85%) is simple: the second buffer absorbs its share of whatever's left after the first. But it's untested by any precedent and may understate correlation: both buffers vanish in the same scenario if the source of loss is the same.

What would kill this, and by when

1. Through 19 Dec 2026, Royco's junior tranche coverage on this market won't fall below 20% (share of total market size, on-chain). If it does, the 34.85% threshold is wrong and the protection maths has to be redone.
2. By 19 Dec 2026, the ROY-ST-savUSD share rate (convertToAssets) will rise at least 1.5% from 0.862705 savUSD (19 Sep). At ≈ 7.1% APY, about 1.8% should accrue over three months. If not, the yield isn't passing through or is leaking somewhere along the way.

What we don't know

• The tranche logic has no published source — how shares are computed and losses split can't be verified. This is the structure's main open risk.
• Whether the market really runs on the current deployment. The "direct tranche" conclusion rests on three indirect proofs and one clean negative; no source says it in words. If the market is on the previous deployment, which Royco itself flags as deprecated "with known accounting limitations", the picture changes.
• The event test didn't fire: the cleanest way to tell a tranche from a vault is the shape of exit events, but there was no flow during nine hours of observation.
• Royco junior tranche over time: the 24% was read on one date; whether it's been shrinking is unknown.
• Nobody knows: who the five signers are (anonymous addresses, no names) and whether Royco's junior tranche would survive a real loss — structures like this haven't been loss-tested in this protocol yet.

Research for information only. Not investment advice.