Demo — not yours · wallet-scoped numbers are a demo wallet · live market data stays live
Borrow against dollars that keep earning.
Carry borrows CDT and routes it through venues in the same motion — the root system of the network. Venue yield flows to lenders, which holds the loan’s cost down, and every closed route feeds the core that funds the borrowing side; neither side extracts from the other. Plain debt, minted to your wallet and deployed nowhere, is Borrow.
USD
One decision. The preset picks collateral, leverage, and the venue mix together — they are the same risk axis — and the confirmation lists every choice it made for you.
Measured routes on other protocols
measured route spreads, including the losing routes
Aug 2026 route cohort, not a count of positions still open today · 12 of 25 priced routes shown · $10,000 examples apply August annualized spreads, not current yield or a forecast · cohort capital was not measured · borrow venue attributed from route semantics · spreads are per dollar, not on leveraged equity
median external net +0.60% / yr per dollar · the Balanced preset carries +19.5% on your equity at 3× — the difference is the loop
Will the venue give it back?
recall behavior when it mattered — evidence for liquidation-time liquidity, not a lender-flight signal
Aave V3money market
fill 100.0% ($412k of $412k, 90d) · slashed 0 · liq recalls 34/34 · no ban
every liquidation recall served in full
your $10k = 2% of everything this venue served in 90d
composition USDe reserve, readable on-chain [high] · withdrawal path synchronous recall inside every withdraw [med] · delivery 34/34 recalls served — history, not a guarantee [low]
Morpho Vault V24626 vault
fill 99.6% ($267k of $268k, 90d) · slashed 0 · liq recalls 41/41 · no ban
one partial fill re-served next block
your $10k = 4% of everything this venue served in 90d
composition lending positions, readable on-chain [high] · withdrawal path buffer self-heals one layer down on withdraw [med] · delivery 41/41 served — history, not a guarantee [low]
sUSDeethena cooldown
fill 87.0% ($132k of $151k, 90d) · slashed 1 · $12k · liq recalls 9/11 · no ban
2 recalls fell inside the cooldown — served after the liquidation had already closed
your $10k = 8% of everything this venue served in 90d
composition staked USDe + silo, readable on-chain [high] · withdrawal path global cooldown; nothing in the exit path starts it [med] · delivery 2 recalls landed inside the cooldown window [low]
PT (fixed maturity)pendle
fill 100.0% ($97k of $97k, 90d) · slashed 0 · liq recalls 6/8 · no ban
2 pre-maturity recalls sold at market: full size, −0.9% price
your $10k = 10% of everything this venue served in 90d
composition fixed-maturity principal token, readable on-chain [high] · withdrawal path pre-maturity exit is a market sale, not a redemption [med] · delivery 2 pre-maturity sales cleared at −0.9% [low]
mock — in production this is reconstructed by an indexer from VaultServed / VaultSlashed / VenueRecalled events and globalVenueBannedUntil; no on-chain aggregates exist · redemptions are 1:1 at peg, so there is no premium to show
Higher Yield Comes With Risks
Staked USDat · 11.53% net vs Morpho Vault V2 · 3.44% net · what you keep after exit costs, over 90 days, at your size, ×10 and ×100
Each line: what a deposit is worth after paying to exit, as % of what you put in, day by day. Green numbers are ahead, red are behind. Hover for the model’s cost range at any day.
Staked USDat · 11.53% net · higher yield, thinner exit
Morpho Vault V2 · 3.44% net · lower yield, deeper exit
your size · $10,000
×10 · $100,000
×100 · $1,000,000
Past ~$590,000 the cheaper venue wins: the higher yield costs more to leave than it pays.
what to take from it
At your size ($10,000), after 90 days you keep +2.74% on Staked USDat vs +0.80% on Morpho Vault V2. Staked USDat wins.
At ×10 ($100,000) it is +2.74% vs +0.80%. Staked USDat wins; the higher yield is eaten by a deeper exit.
The crossover is near $590,000. Below it, chase the yield; above it, pay for the exit.
modelled — route APRs are the measured Aug 2026 table; depth tiers and exit costs are a model, not a measurement, and the band width is the model’s cost range. The venue recorder replaces this with observed withdrawal-ability as history accrues.
Breach, cure window, liquidation
the mechanism, in order
HEALTHY
LTV under your cap
BREACH · 8H TIMER STARTS
0h 00m of 8h
↑ cure anytime in the window: deposit, repay, or price recovery → back to healthy
LIQUIDATION
timer expired uncured · repays to the cap
Two ways in: cross the cap and the timer starts; jump 4% past it and liquidation is immediate, no window.
Liquidation repays to the cap — not the whole loan.
What the venues changed
state changes only — parameter moves and >20% liquidity shifts; drift never appears here
Nothing yet — the recorder logs its first entry when a venue actually changes something.
blind spots unknown: coverage not loaded
observed = witnessed live by the hourly recorder · reconstructed = derived from archive state; real transitions the recorder was not yet running to see · alarm = a failure-pattern flag (danger); its evidence numbers are in the line
Allocation cap activity
Morpho Vault V2 · onchain requests and executions
These requests can change how much a vault may allocate. They do not tell you how much can be withdrawn.
Loading cap requests…
What's being said
raw headlines, newest first — information, not endorsement; we do not summarize or score
Nothing yet — headlines appear once the news fetcher has run for these venues.
Google News RSS · fetched never · membrane offers these venues; headlines are the internet's, not ours