The belt-rank risk framework
Four ranks, one ladder, and the specific evidence that moves an asset up it.
Why a ladder and not a list
A list of approved collateral tells a lender nothing about the distance between the safest and riskiest thing on it. A ladder does. Four ranks, each with a fixed LLTV and a fixed cap, so the position of an asset is the summary of its risk.
White belt
Long tail and memecoins. LLTV 50%, cap $25K, priced from a TWAP only. The buffer is wide because the exit is thin, and the cap is small because the exit is thin.
Green belt
RWAs and mid-cap tokenized equities. LLTV 65%, cap $120K. A priced feed plus a fallback that can disagree. A quarter without an oracle gap is the entry requirement.
Brown belt
Majors and the large tokenized equities. LLTV 77%, cap $400K. Two independent feeds, real venue depth at cap size, and a graded drawdown test the liquidation engine cleared without spilling into another market.
Black belt
USDG, WETH and the proven core. LLTV 86%, cap $1.5M. Reserved for collateral that has been through a cycle on this chain, not one that is expected to.
The parameters are the policy
There is no discretionary override at liquidation time. Whatever the belt says is what the contract does, and changing it is a published parameter change with the same five gates behind it.