Plugin-based macro stress testing over MCP. A person names an event; an LLM agent writes the scenario from a corpus of sources, and the person approves it. Models in levels, from the economy to industries and companies, fill in the paths and reprice positions; each model and asset type is a plugin. Output: loss quantiles, ES95, scenario price.
mcp stress-testing ai-agents brownian-motion expected-shortfall bvar macroeconomic rag scenario-analysis portfolio-risk conditional-forecasting reverse-stress-testing dsh-plugin entropic-tilting ms-var
-
Updated
Nov 9, 2026 - Python