Frequently Asked Questions

StressGen is an AI-powered platform that generates stress testing scenarios for financial institutions. It supports four scenario types — historical, regulatory, hypothetical, and black swan — and aligns with DFAST, EBA, PRA, and Basel III frameworks. It combines macroeconomic data, market signals, news, prediction markets, and regulatory documents to produce context-rooted scenarios.

StressGen uses a multi-agent AI workflow. An economist debate proposes primary shocks across macroeconomic variables, a quant validator computes secondary correlations using copula models, and a consistency checker validates the final scenario. All outputs are grounded in real-time data from 12+ live sources.

StressGen supports four scenario types: Historical (replay past crises like the 2008 GFC or COVID crash), Regulatory (aligned with DFAST, EBA, PRA, and Basel III), Hypothetical (forward-looking scenarios driven by market intelligence and news sentiment), and Black Swan (extreme tail-risk events with severe market impact). Each type can be generated at baseline, adverse, or severely adverse severity levels.

StressGen covers 90+ risk factors across equities, interest rates, FX, commodities, credit spreads, volatility, and macroeconomic indicators. Primary factors (55) are proposed by AI economists, while secondary factors (39) are computed through copula-based correlation modeling to ensure statistical consistency.

StressGen supports four scenario types. Historical scenarios replay past crises (2008 GFC, COVID crash, eurozone debt) with historically-calibrated shocks. Regulatory scenarios are aligned with DFAST, EBA, PRA, and Basel III frameworks using RAG over supervisory guidance. Hypothetical scenarios are forward-looking, driven by market intelligence, news sentiment, and prediction market signals. Black Swan scenarios model extreme tail-risk events with low probability but severe market impact.

StressGen integrates 12+ live data sources including FRED (Federal Reserve Economic Data), Yahoo Finance, ECB, Federal Reserve, Tavily for news and web research, Reddit and Twitter for social sentiment, and Kalshi and Polymarket prediction markets for forward-looking indicators.

StressGen supports four regulatory frameworks: DFAST (Dodd-Frank Act Stress Tests) with all 28 required macroeconomic variables, EBA (European Banking Authority) EU-wide stress tests, PRA (Bank of England Prudential Regulation Authority) stress testing, and Basel III capital adequacy stress testing. Scenarios are aligned with supervisory guidance through RAG analysis over 200+ pages of regulatory documents.

StressGen uses retrieval-augmented generation (RAG) to analyze DFAST supervisory guidance documents. The system extracts required shock parameters, variable specifications, and scenario design rules directly from regulatory PDFs, ensuring generated scenarios meet the framework's quantitative requirements.

Regulatory PDFs are ingested, chunked, and indexed in a vector database. During scenario generation, the RAG agent retrieves relevant passages to extract shock parameters, required variables, and scenario design constraints. This ensures every scenario is grounded in current regulatory guidance.

Three AI economist agents with distinct perspectives propose primary shock values through a structured debate of up to three rounds. Each economist justifies their projections using real-time data. A consensus mechanism synthesizes their proposals into coherent, defensible scenario parameters.

StressGen uses Archimedean copulas (Clayton, Gumbel, Student-t) to compute secondary risk factor shocks that maintain realistic statistical dependencies. When primary shocks are set by economists, copula models propagate consistent secondary effects across credit spreads, volatility indices, and emerging market currencies.

StressGen pulls real-time contract prices from Kalshi and Polymarket to capture market-implied probabilities of economic events. These forward-looking signals complement traditional macro data and news, providing a crowd-sourced perspective on potential stress scenarios and their likelihood.

Context-rooted means every scenario is grounded in real, verifiable data rather than arbitrary assumptions. StressGen synthesizes macroeconomic indicators, market data, news, sentiment, prediction markets, and regulatory documents to ensure scenarios reflect actual economic conditions and emerging risks.

StressGen uses AES-256 encryption at rest and TLS 1.3 in transit. Multi-tenant isolation ensures organizations cannot access each other's data. All personal information is automatically stripped via PII sanitization before reaching any LLM provider. StressGen never sells or shares customer data.

No. StressGen automatically sanitizes all personally identifiable information before any LLM call using a dedicated PII sanitizer. Names, account numbers, and other sensitive fields are stripped or replaced with safe placeholders, ensuring no client data reaches external AI providers.

All data is encrypted with AES-256 at rest and TLS 1.3 in transit. Database connections use SSL with certificate verification. API keys and secrets are stored in encrypted environment variables, never in source code or client-side storage.

The free plan includes 10 scenarios per month, 5 chat messages per day, and up to 5 stored scenarios. You get core context layers: macroeconomic data, web research, market data, news, and prediction markets. No credit card required to get started.

Create a free account — no credit card required. Once signed in, describe a stress scenario in plain language and StressGen generates a full, context-rooted scenario within minutes. You can explore results, adjust parameters, and export data from the dashboard.

StressGen offers a generous free tier with core features so you can evaluate the platform before upgrading. For Pro and Enterprise plans, contact our team for a personalized demo and trial period tailored to your institution's needs.

Pro adds RAG document analysis, full repricing with QuantLib, and higher usage limits (200 scenarios/month). Enterprise includes everything in Pro plus sentiment analysis, custom data integrations, custom pricing libraries, unlimited scenarios, and dedicated support for institutional deployment.