Built by a risk manager who got tired of scenarios that didn't reflect reality.
I spent years as a market risk manager at a tier-1 bank. I know the pain. Scenarios are the most important tool in risk management — and at most banks, they're stale, outdated, and disconnected from reality.
The Problem
A Head of Risk spent a week stress testing the 2024 election. By the time the board saw the results, the world had already moved on.
But that wasn't even the scary part.
His team ran ten stress scenarios every week. Most were broken. Some referenced risk factors that no longer existed. One was still modeling LIBOR exposure — LIBOR was discontinued in 2023.
The rest were historical replays: “What if 2008 happens again?” “What if we see another March 2020?”
The problem with historical scenarios isn't that they're wrong. It's that the world has changed. Regulations changed. Market structure changed. Correlations changed. The leverage that blew up banks in 2008 doesn't sit on balance sheets anymore. A 2008 replay today wouldn't hit the same places.
And the hypothetical scenarios weren't better. Shocks applied to arbitrary risk factors, with arbitrary magnitudes. No one could explain why an oil shock would propagate the way the model said it would — because the model had no mechanism. It assumed a static correlation matrix.
A correlation matrix tells you oil and airlines usually move in opposite directions. It doesn't tell you why. And when you don't know why, you can't distinguish between a supply shock crushing airline margins and a demand boom lifting both oil and travel.
Correlation tells you what moved together. Causality tells you why — and when that relationship breaks.
This is the problem I built StressGen to solve. Most scenario engines are built on correlation. StressGen is built on causality.
StressGen starts from what's happening now — news, policy signals, macro data, market regime — and builds scenarios grounded in real-world transmission. Not: “Based on history, how do airlines react when oil prices double?” But: “If the Strait of Hormuz is blocked, how does that shock propagate to my airline exposures?”
That shock propagates through multiple channels. Each channel operates on its own timeline — hours to weeks. Each has a mechanism. Each can be challenged.
Given what's happening today, what breaks tomorrow?
See What Your Scenarios Are Missing
Schedule a technical demo with our team. We'll walk through scenario generation, the causal engine, and how 10 AI agents work together to stress test your portfolio.
Questions? Get in touch