The two-phase model
- Phase 1 — Instant settlement
- Phase 2 — Async rebalancing
Every user swap settles instantly against the protocol’s liquidity pools. The user receives their destination stablecoin in under one second. At this point, the swap is complete from the user’s perspective.Phase 1 is what partners and end users see — fast, atomic, predictable.
How rebalancing triggers work
The system monitors each pool’s balance relative to its target and responds with increasing urgency:Mild imbalance
The pricing engine adjusts the mid-rate via inventory skew, naturally attracting offsetting flow. No external action needed.
Moderate imbalance
The system coordinates with external market makers via RFQ (Request for Quote) to source needed liquidity. Pool availability is maintained throughout.
Each corridor has independently configured thresholds, allowing the system to respond differently to imbalances in deep G10 corridors versus thinner emerging-market corridors.
Time-of-day awareness
Rebalancing is aware of market hours and trading patterns. The system distinguishes between:- Peak hours — When bilateral flow offsets are likely. Rebalancing is more patient, waiting for natural correction.
- Off-peak hours — When external rebalancing may be needed sooner.
- Holidays — Market holiday calendars for each jurisdiction are factored in.
Self-stabilising design
Multiple mechanisms work together to create a self-stabilising system:Inventory-aware pricing
Attracts offsetting flow before external action is needed.
Tiered thresholds
Escalates response proportionally to imbalance severity.
Yield recall
Provides additional liquidity without relying on external markets.
Cooldown periods
Prevents over-rebalancing by spacing out external actions.