Risk engine

How a limit is derived, and the tiering behind the 97% ceiling.

Spec: README Sections 13-14, 48-49

The risk engine turns holdings into a number: collateral value, maximum borrow, liquidation threshold and health factor. It is intended to be deterministic and independently testable — given the same holdings and prices it must always produce the same limit.

LTV by risk tier

Tier	Examples	Max LTV	Liquidation	To bad debt
Tier 1 — Very liquid	SPY, liquid ETFs	97%	98%	2 pts
Tier 1 — Large cap	AAPL, MSFT, AMZN	96%	98%	2 pts
Tier 2 — High volatility	NVDA	94-95%	97%	3 pts
Tier 3 — Very high volatility	TSLA	92-94%	96-97%	3-4 pts
Tier 4 — Lower liquidity	Small-cap, thinly traded	85-90%	92-95%	5-8 pts

Per $100 of Tier 1 collateral

$100 Collateral
      │
      ├── $97 Maximum Borrow
      │
      ├── 98% LTV → Liquidation Begins
      │
      └── 100% LTV → Insolvency / Bad Debt

Health factor

Health Factor = Liquidation Value / Debt

Above 1.00 the position is healthy; below 1.00 it can be liquidated. Liquidation value is the sum of each holding valued at its own liquidation threshold, so a portfolio of mixed tiers produces a blended figure rather than a single asset’s number.

Inputs to the collateral factor

Realised and implied volatility
Liquidity and liquidation depth
Market capitalisation and trading volume
Oracle quality and update latency
Trading status — halts, market hours
Historical drawdown and correlation