The advisory work runs on models. These are those models, built as software and handed to the people who have to defend the answer in the room. One reads every US bank and credit-union filing, the other reads states. Two subjects, one standard: from any number on the screen you can get back to where it came from.
Credit risk for banks and credit unions
Every US bank and credit union, every quarter of bank filings back to 2008, and an assumption-driven modelling layer on top. You set the deposit betas, the runoff rates and the growth plan; the models run against your own filed numbers and the board pack prints what the screen says.
Country and geopolitical risk
The same discipline turned on states. Covered countries scored on weighted pillars, a live event monitor running underneath, and the balance-of-power, sanctions and trade layers that decide whether a sovereign balance sheet means what it says. Scenarios are structured rather than narrated, and forecasts are scored once the answer is known.
When a source does not report a line, it is shown as not reported, and so is every figure computed from it. Filling the hole with a zero is how a model ends up flattering the subject it should worry about.
Reported, derived, or assumed, stated beside the figure and carried into anything printed. A reconstructed value is useful; a reconstructed value presented as reported is not.
A measure is computed in exactly one place and reused: a ratio, a pillar score, a rank. Two screens that disagree about the same number destroy more trust than a missing screen ever would.
Neither product is self-serve. Access is granted per organisation and invoiced, which is why there is no sign-up button anywhere on this site. Lynx is open to new institutions now; Renard is in beta and taking a limited number. Tell us who you are and what you need it for, and we reply either way.
Say which of the two you are after, or describe the problem and we will tell you which one fits, if either does.