Parametric Cover Pricing Practice: Trigger Selection, Loss Ratio, Capital Efficiency

Date: 2026-10-19T03:00:00Z  |  Cycle: autonomous research cycle 2026-10-19  |  Topic: Backlog P2 Zero-Storage research-todo-017-006 — parametric-cover pricing practice (unblocks s9-todo-006)  |  Revisit: 2026-10-26

Summary

s9-todo-006 "Build parametric cover quote engine" (Stream 9, §03.6, MEDIUM, PENDING) waits on pricing mechanics before any code: how to pick a trigger, how to derive a premium from a modelled loss ratio, how much capital to hold, and whether calling the product "cover" instead of "insurance" matters under MiCA. Key result: price premium = (expected-loss frequency × severity) × risk loading + cost of capital, quoted only when a modelled loss ratio exists; triggers must be objective and onchain-verifiable (depeg, failed liquidation, frozen withdrawals); capital is sized by a solvency ratio (target CAR 1.5–2.0, leverage capped 1.5–5x) over a first-loss buffer plus staked collateral; cover is explicitly not insurance — wording must avoid indemnity/guarantee language, and MiCA treats an insurance policy only as one permitted prudential safeguard, not authority to underwrite regulated insurance. Finding #008 covered Firelight's model; this finding covers the pricing practice that lets `coverage-underwriting.js` actually quote. Unblocks s9-todo-006.

Key Findings

• Trigger selection (Firelight Cover Scope, primary): covered events are binary and objective — bad-debt share greater of 5% of position or $50,000; liquidations recovering under 80% of fair realizable value; a pegged asset trading more than 5% off peg for over 4 consecutive hours (VWAP across 3+ liquid venues); position not withdrawable for over 72 hours, or returning under 90% of fair value. Detection by an independent threat-detection vendor recorded onchain; loss assessed by the incident reporter; validated by a 3-of-5 Risk Consortium; a 24-hour stabilization period applies to economic-risk events; one cause = one recovery; duty to mitigate applies. This matches EIOPA's parametric definition: a lump-sum payment on an objective predefined trigger, not loss indemnification.
• Loss-ratio pricing (SOA + Firelight Risk Pricing Framework): expectation-principle premium = (1 + loading) × expected loss, where expected loss = trigger frequency × average severity; the standard-deviation principle adds loading × standard deviation for tail risk. Firelight prices each market from two independent views — economic (bad debt, collateral/graph structure) and technical (contract bugs, oracle fragility, upgrades) — and publishes a base premium rate plus available limit per market to the Cover Registry on a 30-day cadence, fixed within the period. Premium reflects expected claim cost plus cost of backing capital.
• Capital efficiency (Firelight Capital Adequacy): Capital Adequacy Ratio = Available Capital ÷ Solvency Capital Requirement; target band 1.5–2.0 steady state, 1.75–2.0 at launch. Effective leverage bounded 1.5x–5x (target operating 2.5–4x). Five CAR bands drive automatic action: ≥2.0 strong; 1.5–1.99 adequate; 1.2–1.49 enhanced monitoring and restricted new coverage; 1.0–1.19 suspend new coverage except renewals; below 1.0 critical, reduce renewals and restore capital. Available capital = first-loss buffer + market value of staked assets; capacity = buffer + staked × leverage − active coverage − renewal commitments.
• Wording and MiCA risk: Firelight states cover is not insurance, an indemnity, a guarantee, or a regulated financial service — no insurer, broker, or financial-institution role. Nexus Mutual cover terms show the market-standard mechanics to mirror: deductible subtracted from loss, claim amount capped at cover amount, claims valued via a reputable crypto pricing source, claim window from 14 days after the event up to 35 days after cover period ends. MiCA Article 67 lets a crypto-asset service provider meet prudential safeguards with an insurance policy from an authorised EU undertaking (≥1 year term, ≥90-day cancellation notice, third-party) — that is a safeguard option, not a licence to sell regulated insurance. EIOPA notes smart-contract parametric models may fall under MiCA as utility tokens; insurance intermediaries selling them need compliance.
• Stream 9 zero-storage fit: premium and trigger evaluation are computed on request from oracle feeds (Pyth/Chainlink) and onchain registry reads; nothing is persisted beyond single-period state, no custody, no owned datasets — consistent with §06 and finding #011's free-tier/stateless-query limits.

Recommendations

Impact on Streams / Todos

• Stream 9 (Zero-storage): unblocks s9-todo-006 (parametric cover quote engine, MEDIUM — research-todo-017-006 now complete, ready to implement). Builds on finding #008 (Firelight model), #011 (stateless query limits), and #015 (backstop-depth safety signal).
• Stream 1 (DeFi): stablecoin yield quotes gain cover-cost context; do not quote the extra reward rate as yield without backstop depth (finding #015).
• Infra: revisit Firelight pricing/cover docs at 30-day rate-refresh cadence; next revisit 2026-10-26 (high value, 7 days).

Sources