Medici Protocol

Borrow against BTC.
No liquidation price exists.

Split BTC into two tokens — one stable-ish, one leveraged. Use the stable side as collateral that mathematically cannot be liquidated.

How the mechanism works API Reference

The problem with BTC lending today

Every BTC lending protocol — Aave, Maker, Compound, and every perp DEX — shares the same failure mode: liquidation.

Liquidation cascades

One large liquidation depresses the price, triggering more liquidations, depressing the price further. MakerDAO's Black Thursday (March 2020) saw $8M in collateral liquidated for $0 due to network congestion. The mechanism itself amplifies drawdowns.

Oracle dependency

Liquidation engines need accurate, real-time prices. Oracle lag — even seconds — creates arbitrage windows. Oracle manipulation (Mango Markets, $116M) exploits the gap between reported price and true price. The liquidation trigger is a single point of failure.

Cliff-shaped risk

Your position is worth 100% until it hits a threshold, then it's worth whatever a forced auction returns — minus a liquidation penalty, minus slippage, minus MEV extraction. The risk profile is binary: fine one block, wiped out the next.

How Medici fixes this

Split 1 BTC into two tokens — P (Protected) and N (Amplified) — whose payoffs always sum back to the collateral.

P Token — the collateral

  • Tracks BTC with bounded roll-execution cost (~1.2%/yr at 10bps slippage). Gross theta collected: +0.57%/yr (real vol). Net roughly break-even; outperformance depends on execution quality.
  • Value at maturity = S/x, fully computable at every price before entering
  • Collects theta — P holders are option sellers, paid for accepting upside risk
  • No forced liquidation — no margin calls, no forced sales, no bad debt. P's value can decline above the strike (continuous, fully computable, never a cliff).

N Token — the leverage

  • Leveraged BTC upside — uncapped call option
  • No funding rate, no forced liquidation, no ADL queue
  • Pays theta to P holders
  • Expiry-bound, strike-bound — not a perpetual replacement, but a defined-risk alternative

P + N ≡ 1 BTC · always

That single invariant means the position is never under-collateralized. "No forced liquidation" is a theorem, not a risk parameter. It is not a claim that principal is protected — P's value declines continuously above the strike. That decline is fully computable at every price and is the source of P's yield.

The lending math

At maturity with BTC price x and strike S:

ConditionP receives (BTC)N receives (BTC)
S ≥ x (BTC at or below strike)1.000.00
S < x (BTC above strike)S/x1 − S/x

A lender underwriting against P isn't modeling tail risk on volatile collateral. They're lending against a token whose minimum value is mathematically bounded. The liquidation cascade that kills every other lending protocol simply cannot happen — the invariant forbids it.

ScenarioP value at maturitySafe LTV
BTC/USD at strike or below1.00~90%
BTC/USD 2× strike0.50~45%
BTC/USD 5× strike0.20~18%

LTV = loan-to-value against P's known minimum. Compare to 70–80% LTV on volatile collateral with liquidation risk.

What this costs

No liquidation isn't free. Being honest about the costs builds credibility.

Roll-execution cost: ~1.2%/yr

At 10bps slippage with ~12 rolls/year, the execution cost is ~1.2%/yr. Gross theta collected by P is +0.57%/yr (real vol). Net is roughly break-even. This is not a yield guarantee — it's the cost of maintaining the position. Comparable to ETF expense ratios, and far cheaper than a liquidation penalty.

Fixed term

This is a term-structured product, not an open-ended loan. You trade liquidation risk for a known maturity date. Rolls are possible (atomic, auction-based), but liquidity at roll time isn't guaranteed.

Sale of a tranche

You're not borrowing against collateral — you're selling the P tranche to a buyer who wants its theta. Different capital efficiency, different UX, different risk profile from a revolving credit line.

Why this beats alternatives

Medici P TokenLending protocol (Aave/Maker)Perpetual swap
Forced liquidation riskNone — no margin calls, no forced sales, no bad debtYes — price-basedYes — funding + margin
Oracle dependencySettlement only (hours/days)Real-time (seconds)Real-time + funding rate
CostRoll cost ~1.2%/yr at 10bps; gross theta +0.57%/yr; net ~break-evenVariable borrow rateVariable funding rate
TermFixed, known in advanceOpen-ended (until liquidation)Open-ended
Capital efficiencyModerate (split + sell P)Moderate–high (LTV-dependent)High (leverage)
Privacy-preservingYes (Canton, Midnight)No (public chain)No (public chain)

The N side: Amplified vs a perps-DEX

The comparison above is the P story — index-tracking collateral for the lending audience. The N token has its own natural alternative: the perpetual future. It is a different instrument for directional exposure — defined-risk convexity, not a margined perpetual — so the contrast is structural, not "a cheaper perp." Perps are now explicitly CFTC-regulated onshore; the honest version below is the credible one.

Perps-DEXMedici (Amplified)
Forced closureMargin call → liquidation engine → backstop vaultNone: P + N = 1 is a settlement identity — no under-collateralized position can be seized
Intraday wickA wick through your liquidation price ends the position — even if price recovers an hour laterOnly the settlement price at maturity matters; a wick can never reach you
Cost of carryFunding rate — variable, open-ended, charged periodicallyTheta — the premium, prepaid once at entry; fixed and known, nothing accrues against you later
Maximum lossFull margin, plus socialized loss (ADL) if the backstop vault is squeezedCapped at the premium you pay as the option buyer — known before you enter
Where perps win
honest concession
Deep, continuous liquidity: exit any second at tight spreads, and far higher leverage (10–50×)~2× exposure, fully collateralized; a clean exit means holding both tokens or a thinner secondary market

Vocabulary is deliberate: Medici is described only in options terms (option buyer, premium, theta, strike, maturity, settlement identity). Margin, funding, liquidation and borrowed leverage belong to the perps column — the contrast is between two instruments, not a claim to be one.

Use cases

BTC treasury hedging

Miners and corporate treasuries holding BTC can split their holdings: sell P to raise cash at a known minimum, keep N for upside — with no liquidation to blow up the operation mid-drawdown.

DAO runway protection

A DAO with treasury in ETH or a native token can split into P (known-dollar runway, collects theta) and N (sell for cash, or keep for upside). No open-market sale, no price impact, no "team is dumping" signal.

Defined-risk leverage

N tokens are uncapped call options — leveraged upside, no funding rate, no liquidation price, bounded loss. A direct answer to what traders hate most about perps.

Where it stands

Working MVP on Canton (dev-deployed, not mainnet). 36 DAML templates, 12-agent fleet, 9-year backtest (Hypercall, Aug 2017–Jun 2026). The mechanism is proven. The port to Midnight is the next step.

Based on Vitalik Buterin's "Building Index-Tracking Assets on Top of Options Instead of Debt" (ethresear.ch/t/25036). Backtest: Hypercall (insights.hypercall.xyz/options-not-debt-2026-06-11). Live Solidity reference: mmchougule deployment on Base.