Yield Lab
We judge live stablecoin yield against the RWA floor (~3.4%). Every basis point of spread must be explained by a measurable, accepted risk — or it is refused. We publish the refusals too: that discipline is the product.
How to read this
These are research judgements (evidence level L2 — from public sources as of 2026-07-02), not
live returns, an offer, or investment advice. "ADVANCE → paper" means an advisory paper test with
no capital, separate from our go-live track. Numbers drift; the source of truth is
docs/decision_index.md + docs/underwriting_rubric.md in the public repo.
The fundable candidates (ADVANCE/WATCH) now record an advisory paper track in the
Strategy Lab (agent com.spa.strategy_lab_paper, hourly) — no capital,
separate from the go-live track. Forward curves accrue over time; this is a simulation at the sourced rate, not realized yield.
Decisions on live data
PT-sUSDe
Pendle fixed carry
Fixed-to-maturity removes the funding-flip tail; USDe solvency stress-validated (Oct-2025 crash: overcollateralized throughout). Capacity ~single-digit $M/maturity.
PT-USDe
Pendle fixed carry
Same fundable structure, but same Ethena underlying as PT-sUSDe → shares one cap, not extra diversification (~70% of Pendle TVL is Ethena).
Maple syrupUSDC
Institutional credit
Bounded credit — loans overcollateralized 120–170% at qualified custody (Anchorage/BitGo/Copper), ~3yr zero-loss. DD-gated (v1 lost $50M in 2022).
Centrifuge DROP
RWA senior tranche
Senior tranche bounded by a real junior (TIN) first-loss buffer + real RWA cash flows. Per-pool off-chain DD required (buffer depth, asset quality).
Aave V3 USDC
Blue-chip lending
The safest, deepest DeFi lending pays ≈ the floor — no spread to underwrite. Hold the T-bill floor instead. Proof that spread must be earned by risk.
Ethena sUSDe (spot)
Funding carry
Floating funding carry is an unbounded tail (funding-flip, CEX-counterparty, ~1.1% reserve). Thin spread today does not pay for the tail.
Curve/Convex LP
Stable LP + emissions
The attractive part is CRV/CVX token emissions — a subsidy, not risk-comp. Strip emissions → base fees ≈ floor-parity. Value what you keep if rewards → 0.
Goldfinch Senior
Uncollat. EM credit
The tail fired at scale — protocol formally winding down after ~$50M of defaults stranded depositors ~3 years. A "senior" label does not bound a thin first-loss buffer.
Resolv RLP
First-loss tranche
The yield is payment to absorb first loss with self-balancing leverage — and the tail already fired: a 2026 mint exploit (~$25M extracted, USDe −39% depeg, TVL $400M→$9M).
The non-Ethena ladder
The real 8-12% in stablecoins is structurally dominated by Ethena (funding-carry). You diversify by asset class, not by chain (hopping to Base just re-routes to Ethena). Here is the honest ladder without Ethena concentration:
The pick-two rule
You can't have all four at once — 8-12%, diversified, non-Ethena, and at scale. A diversified non-Ethena book
(40% floor / 30% overcollat / 20% Maple / 10% Centrifuge) blends to ≈4.75% — bounded,
~135 bps over the floor, but not 8-12%. To reach 8-12% without Ethena you must
concentrate in higher-risk credit. Diversifying away from Ethena is the right risk decision, and it
costs yield. That stated trade-off is the product. Full analysis: docs/non_ethena_ladder.md.
The one principle
Yield is inverse to fundability. The biggest headline number (Resolv 20–30%) drew the hardest refusal — its tail already fired. The smallest explained spreads (~160–780 bps) are the only fundable ones. Spread is bought with accepted risk; it is never free.