Finance.
Banks, settlement, and sovereign credit — interconnection as a risk axis.
Banking, settlement, and sovereign-credit networks survive on assumed liquidity that disappears under stress. The counterparty graph is monitored bilaterally; the network amplification — second-order exposure through the rest of the book — is hard to size and rarely priced into limits or capital.
Standard stress tests run firm-by-firm or scenario-by-scenario. They miss the network-amplification term that is the actual mechanism of past crises: 2008's repo run, the 2010 sovereign loop, the 2023 regional-bank cascade. Each was a graph event the firm-level tests didn't see in advance.
Sovereign credit is the same problem at country level. A single political event can re-price an entire sovereign curve overnight, and the structural setup that made it that fragile wasn't visible in the ratings.
Manifold ingests the supervised perimeter (lending exposures, derivatives, settlement flows, payment-rail dependencies) and the sovereign perimeter (cross-border banking claims, FX reserves, debt structure). Edges encode bilateral exposure, settlement flow, and rail dependency.
PEARL runs counterfactuals on named stress scenarios (this central counterparty is impaired; this rail is unavailable for 24 hours; this sovereign loses access). PARETO simulates the resulting cascade across the supervised graph. SPIRTES surfaces structural dependencies the supervisory data implies but isn't called out.
Bilateral exposure is monitored counterparty-by-counterparty. The network amplification — second-order exposure through the rest of the book — is hard to size and never priced into limits.
Graph-based exposure on lending, derivatives, and settlement networks. ΩF per counterparty including indirect path risk.
Standard stress tests run firm-by-firm. They miss the network-amplification term that is the actual mechanism of past crises.
Cascade simulation across the supervised graph. Scenario library mapped to identifiable systemic-stress mechanics.
Country ratings smear over real fragility. A single political event can re-price an entire sovereign curve overnight, and the structural setup wasn't visible beforehand.
Jurisdictional hazard + cascade load on the sovereign-finance subgraph. Counterfactual on named regime-change scenarios.
Run Manifold on a finance graph.
Trial accounts come pre-loaded with a curated dataset. Or request an invite to bring your own graph.
