Stress Scenario Types — DFAST, EBA & Hypothetical

What Are Stress Scenarios?

A stress scenario models a hypothetical economic event and its projected impact on financial markets. What happens to equities, interest rates, currencies, and commodities if a specific shock occurs?

StressGen generates a complete table of risk factor shocks for each scenario — 90+ factors across 8 asset classes — in under five minutes. Every shock comes with full provenance: which regulatory document informed the parameters, which live data feeds provided the baseline, and which statistical model propagated secondary effects.

Traditional scenario engines apply shocks based on static correlation matrices. StressGen starts from what's happening now — news, policy signals, macro data, market regime — and builds scenarios grounded in causal transmission. Not “how did airlines react last time oil spiked?” but “if the Strait of Hormuz is blocked today, how does that shock propagate to airline exposures given current conditions?”

How It Works

Every scenario passes through six layers of context before it reaches your desk:

• Regulatory context — RAG retrieval over ingested DFAST, EBA, and PRA guidance documents extracts shock parameters and severity benchmarks directly from regulatory PDFs.
• Market context — Live data from FRED, Yahoo Finance, ECB, and 12+ sources provides the current baseline — credit spreads, yield curves, equity indices, and commodities.
• Market intelligence — Hundreds of news articles, SEC filings, prediction markets (Kalshi, Polymarket), and social signals are synthesized into forward-looking context. The result is a summary of summaries — not raw noise, but distilled signal.
• Statistical context — C-Vine copulas propagate primary shocks into secondary effects with statistical coherence. Every cross-asset dependency is computed, not assumed.
• Historical context — 20+ years of risk factor data calibrates every shock against real historical distributions and precedent.
• Likelihood intelligence — A quantitative scoring system compares each scenario against current market conditions, producing a 0–100 likelihood score.

10+ specialized AI agents execute this pipeline in parallel — from regulatory extraction through economist debate to copula propagation — delivering a fully attributed scenario in minutes, not days.

Scenario Types

StressGen supports four scenario types. Each uses the same six-layer pipeline, but with different emphasis and calibration:

Historical

Replay past crises — 2008 GFC, 2020 COVID crash, eurozone debt crisis, dot-com bust — with shocks calibrated against actual market impacts across all risk factors. But StressGen doesn't just replay old correlations. It re-grounds each historical event in today's market structure, regulations, and leverage profiles, so the shocks reflect how that crisis would propagate now.

Use when: You need a regulatory-accepted benchmark, want to stress test against a known crisis, or need to show how your portfolio would have performed during a specific historical event.

Regulatory

Aligned with DFAST, EBA, PRA, and Basel III supervisory frameworks. StressGen uses RAG (retrieval-augmented generation) to extract exact shock parameters, risk factor coverage requirements, and severity benchmarks directly from regulatory guidance documents — not from training data.

Use when:You're preparing a regulatory submission (CCAR/DFAST, EBA stress test) and need scenarios that meet specific supervisory requirements for risk factor coverage and severity calibration.

Hypothetical

Forward-looking scenarios grounded in current market intelligence. You define the triggering event — the AI synthesizes hundreds of signals from live news, prediction markets, macro data, and SEC filings to generate shocks that reflect today's conditions, not historical averages.

Use when:You want to model events that haven't happened yet — an emerging geopolitical risk, a policy shift, or a sector-specific disruption — and understand how it would ripple through your portfolio given what's happening now.

Black Swan

Extreme tail-risk events with low probability but severe market impact — pandemic escalation, sovereign default, unexpected geopolitical escalation, or systemic financial contagion. Black Swan scenarios push shock magnitudes to the boundaries of historical distributions and stress test whether your portfolio survives conditions that models typically don't cover.

Use when:You need to test portfolio resilience under extreme conditions, satisfy board-level “worst case” requirements, or explore scenarios beyond the range of typical stress tests.

Creating a Scenario

Every scenario starts with a description of the economic event you want to model. Write it in plain English — the AI will interpret your intent and determine the appropriate shocks.

Good scenario descriptions include:

• The triggering event (e.g., a central bank rate decision, geopolitical conflict, pandemic)
• Expected severity (mild correction vs. severe crisis)
• Time horizon and affected regions or sectors
• Any specific risk factors you want emphasized

Scenario Workflow

Every scenario follows a lifecycle with clear statuses:

Editing & Versioning

After generating a scenario, you can refine it:

• Edit description — Update the scenario narrative and regenerate shocks with the revised context.
• Adjust data sources — Toggle different data feeds on or off to see how they influence the results.
• Version history — Each regeneration creates a new version so you can track changes over time.

Comparing Scenarios

Use the comparison feature to view multiple scenarios side by side. This is especially useful for:

• Comparing regulatory baseline vs. severely adverse scenarios
• Evaluating how different trigger events affect the same risk factors
• Reviewing how scenario shocks change when different data sources are enabled

Interpreting Results

Each generated scenario includes:

• Shock table — The complete list of risk factor impacts organized by asset class (equity, rates, FX, commodities, credit, volatility, macro, inflation). Positive values indicate increases; negative values indicate declines.
• Narrative summary — A plain-language explanation of the scenario's economic logic — the causal chain from trigger event through transmission channels to each shock magnitude.
• Likelihood score — A 0–100 quantitative score comparing how probable this scenario is given current market conditions, computed across four signal dimensions with confidence bands.
• Supporting evidence — The real-world data points that informed the AI's decisions: economic indicators from FRED, news articles, SEC filings, prediction market signals, and regulatory precedent.
• Historical narrative — For historical scenarios, a data-grounded narrative covering the crisis timeline, peak-to-trough equity impact, policy responses, and transmission mechanisms.

Exporting & Sharing

Scenarios can be exported for downstream systems or regulatory submissions:

• CSV / Excel — Structured shock tables ready for import into risk management platforms, VaR engines, or internal models.
• PDF report — Presentation-ready summary with shock table, narrative, methodology notes, and supporting evidence for stakeholder or board review.
• Public sharing — Generate a shareable link to a read-only view of the scenario, with OG meta tags for previews on Slack, email, or social media.