A working model of an AUD stablecoin to USDC swap with the swap-back rate fixed in advance. It prices the forward, sizes the margin buffer, and draws its USD yield from optimised portfolios of regulated and on-chain providers. Rates on load are the latest daily reference rates; other inputs are illustrative assumptions, not quotes.
Daily reference rates, not live dealing prices. The model uses them until you change an input: AUD/USD sets spot, RBA cash is a proxy for the AUD rate the provider earns, SOFR for the provider's USD opportunity cost, and the USD DeFi portfolio sets DeFi USD yield and expected loss. "Reset to market rates" restores them.
Optimised portfolios of USD stablecoin yield sources with TVL of US$20m or more, snapshot of 23 October 2026. The optimiser maximises yield after expected loss, subject to liquidity, concentration and pool-share limits. "Use" copies the portfolio's model USD APY into DeFi USD yield and its expected loss plus venue costs into the model.
Conservative: risk tiers 1 and 2 only, at most 15% in any one position, no private credit or Ethena, and at least 30% redeemable within a day and 95% within 30 days. Balanced: tiers up to 3, at most 20% per position, up to 20% private credit and 10% Ethena, and 90% redeemable within 30 days. Target-seeking: tiers up to 4, at most 25% per position, up to 40% private credit and 15% Ethena, and 80% redeemable within 30 days. Stretch: target-seeking limits with private credit up to 60% and Ethena up to 20%; it shows what more concentration buys and is not a proposed mandate. Every appetite keeps 10% redeemable the same day and excludes tier 5.
Stressed liquidity assumes lending-pool and vault same-day liquidity goes to zero (as in the April 2026 Aave stall) and savings-rate same-day capacity halves. Margin calls in the first week of a stress would have to be met mostly from the margin buffer.
| Component | % p.a. |
|---|
The hedge covers principal. USD yield accrues unhedged and is converted at the spot rate when paid, so it carries FX risk on the yield amount only. Rebalancing the notional each day keeps that small.
| AUD move | Spot | VM owed (USD) | Buffer cover |
|---|
When AUD falls, the investor owes variation margin to the provider. When AUD rises, the provider owes, and the investor relies on the provider's segregated AUD deposit. The worst AUD falls tend to coincide with crypto stress and DeFi withdrawal gates.
Carry in % p.a. by AUD rate earned (rows) and provider's USD opportunity cost (columns), including the provider spread. Positive cells mean the hedge adds return. The outlined cell is closest to the current inputs.
156 USD stablecoin yield sources with TVL of US$20m or more; snapshot of 23 October 2026. Sources: DefiLlama, Morpho, Fluid and Jupiter APIs, rwa.xyz and issuer pages. Yields move daily. Grey rows are not used by the optimiser; the reason is shown. Rows refresh from DefiLlama, Morpho, Fluid and Jupiter every day and on update; ten hand-collected rows are reviewed monthly. A row dated more than 14 days ago is marked stale, and more than 30 days ago it is left out of the optimiser. Risk flags and default assumptions are judgement and change only at review.
Data freshness: loading…
| Provider | Product | Category | Chain | Asset | TVL US$m | Base % | Reward % | Total APY % | Model APY % | EL-adj % | Tier | Days normal | Days stressed | Same-day US$m | Risk flags | Optimiser | Data | Australian access | Note |
|---|
USD stablecoin venues of US$20m or more yielding at least 1% that are not in the dataset, from a weekly DefiLlama scan. They are not used by the optimiser. Adding one means setting its risk flags in the dataset, which is a judgement call.
Model APY is the spot APY with token rewards counted at 50%, and the 30-day mean substituted where spot exceeds it by more than 1 point. EL-adj is model APY minus expected loss (probability of default × loss given default by tier, a judgement) minus venue cost. Lending-market TVL is total supplied, not DefiLlama's free-liquidity figure. Probabilities of default, loss rates, haircuts, costs and caps are judgement assumptions.