BTR, the inventory-based AMM
BTR is an Adaptive Inventory Market Maker (AIMM): a multi-asset automated market maker that quotes against an external signed mark instead of an internal bonding curve. Liquidity is shared across every asset in a core rather than locked into pairs, deposits are single-sided and fungible, and there is no range to set or rebalance.
Because the mark comes from outside the pool, the price does not have to be discovered by arbitrageurs walking a curve. Quotes bend with inventory: the side that restores balance gets price improvement, the side that pushes it further pays a spread that widens with imbalance, volatility and mark staleness. That is what keeps loss-versus-rebalancing from being paid out of liquidity providers’ positions.
The pricing curve and the spread logic are both on chain and verifiable, which is the part proprietary oracle AMMs keep closed.
How pricing works
A BTR pool does not discover price by walking a bonding curve. It quotes around a mark signed by NX Rates keepers and verified on chain, then adjusts that quote for the pool’s own inventory. The further a leg drifts from its target coverage, the wider the quote gets for anyone pushing it further out, and the tighter it gets for anyone bringing it back.
Depth is shaped by a quartic I-spline stored per pool, so liquidity concentrates where the asset actually trades and moves as the mark moves. Nobody sets a range and nobody rebalances one.
What liquidity providers get
Deposits are single-sided and fungible: one asset in, a share of that leg’s liability out. There is no pair to assemble and no position to manage. Returns come from the spread that flow pays, not from token emissions, so the reported yield is yield that was actually earned and it does not end when a programme does.
Coverage-based protection bounds impermanent loss per leg, and the withdrawal path prices the imbalance a withdrawal creates rather than socialising it across the remaining providers.
Why the oracle matters
Pricing against an external mark is what removes loss-versus-rebalancing as a structural cost. It also introduces staleness risk, so staleness is priced: the spread carries a premium proportional to volatility times the square root of the time since the mark was signed. That is what makes a push-oracle AMM safe to run between pushes.
The signer set, the staleness bounds and the verification path are all on chain and independently checkable from the oracle page.
Status and where to read more
BTR is live on Arc testnet. The documentation covers the pricing model, the oracle wire format, access control and emergency powers, the risk parameters, and the deployed contract addresses. There is a public read-only API and an OpenAPI specification for programmatic access, and a bug bounty for anything you break.