Purpose
Convert a described risk scenario into a defensible financial impact range — with expected loss, tail loss, and scenario spread — using Value-at-Risk logic, sensitivity analysis, and stress modeling. Turns "this feels risky" into numbers a CFO can take to the board or use in a reserve decision.
When to Use
Invoke this skill when the user asks to:
- Size the potential financial loss from a specific exposure or event
- Estimate Value-at-Risk on a position, portfolio, or currency exposure
- Run a scenario or stress test on a market, credit, or liquidity risk
- Quantify counterparty or concentration exposure for a credit decision
- Convert a qualitative risk-register entry into a dollar impact range
- Justify a loss reserve, hedge budget, or capital buffer with numbers
Inputs Required
Before starting, collect:
- Risk type: Market / credit / liquidity (or a mix — identify which drivers apply)
- Exposure base: The notional or dollar amount at risk (position size, receivable balance, funding gap)
- Risk driver and range: What moves (rate, FX, price, default probability, drawdown) and its plausible move range
- Time horizon: The period over which the loss could materialize (1-day, 10-day, quarter, year)
- Confidence level: Desired VaR confidence (default 95%; also report 99% for tail)
- Existing mitigants: Hedges, collateral, covenants, credit insurance, committed facilities in place
- Volatility or historical data: Standard deviation, historical loss rates, or recovery rates if available
If volatility or historical data is missing, use published proxies for the asset class or driver and label the assumption explicitly.
Instructions
Step 1: Classify the Risk and Select the Method
Match the risk type to the primary quantification method:
- Market risk (rate, FX, price, equity, commodity moves): Parametric VaR as primary, sensitivity (DV01/delta) as supporting view
- Credit risk (default, downgrade, concentration): Expected Loss = Exposure × PD × LGD, plus concentration stress
- Liquidity risk (funding gap, forced-sale, rollover failure): Cash-flow gap analysis plus liquidation-haircut modeling State which method drives the headline number and which serve as cross-checks. If multiple drivers apply, quantify each separately, then combine (note correlation assumptions if aggregating).
Step 2: Establish the Baseline and Drivers
Lay out the quantification inputs explicitly before calculating:
- Restate the exposure base and time horizon in one line
- State the risk driver, its baseline value, and the move range (in bps, %, or price units)
- Convert horizon: scale volatility by √time when moving between daily and multi-day horizons
- Document every assumed input (volatility, PD, LGD, recovery, haircut) with source or proxy label
Step 3: Calculate Value-at-Risk / Expected Loss
Run the primary calculation and show the formula with numbers substituted:
Market risk (Parametric VaR)
- VaR = Exposure × z-score × σ × √(horizon)
- z-scores: 1.645 (95%), 2.326 (99%)
- Report both 95% and 99% VaR
Credit risk (Expected + Unexpected Loss)
- Expected Loss = Exposure at Default × Probability of Default × Loss Given Default (LGD = 1 − recovery rate)
- Unexpected Loss = the tail case using stressed PD/LGD (e.g., recession-level defaults)
Liquidity risk
- Funding gap = outflows − (inflows + committed liquidity) over the horizon
- Forced-sale loss = illiquid assets × liquidation haircut
Show the arithmetic — never present a final figure without the calculation path.
Step 4: Sensitivity Analysis
Show how the impact scales with the driver, so the reader sees the shape of the risk:
- Build a small table: driver move (e.g., +50bps, +100bps, +200bps) → resulting loss
- Identify the single most impactful variable (the one the loss is most sensitive to)
- Note any non-linearity (convexity, gap risk, correlation breakdown under stress)
Step 5: Three-Scenario Quantification
Present the impact as a range across three named scenarios:
- Base case: Most likely outcome — the expected loss figure
- Adverse case: 95th-percentile move — the VaR figure
- Severe case: Tail / stress move — 99% VaR or a defined stress event For each, state the driver assumption and the resulting dollar loss.
Step 6: Apply Mitigants and Net the Exposure
Adjust the gross figures for existing controls:
- Net each scenario for hedges, collateral, insurance, or committed facilities
- Show gross loss → mitigation offset → net residual loss
- Flag any basis risk or mitigant gaps (e.g., hedge covers 70% of notional, collateral revalues under stress)
Step 7: Summarize and Recommend
Deliver the headline range and one clear action:
- State the net residual loss range in a single sentence (base to severe)
- Recommend a specific action tied to the number: reserve level, hedge quantity, exposure cap, or facility size
- Note the largest unquantified uncertainty so the reader knows the estimate's limits
Output Format
- Structure: Method classification → input baseline → VaR/EL calculation with shown arithmetic → sensitivity table → three-scenario range → mitigation netting → summary recommendation
- Length: 700–1,000 words
- Format: Structured document with calculation blocks, one sensitivity table, and a scenario table; prose kept tight
- Sections: Risk Classification & Method → Baseline & Drivers → Core Calculation → Sensitivity Analysis → Three-Scenario Range → Mitigation-Adjusted View → Summary & Recommendation
- Tone: Risk director briefing a CFO — quantitative, assumption-explicit, no hedging language beyond what the numbers require
Quality Checklist
Before delivering output, verify:
- [ ] The primary calculation shows the formula with actual numbers substituted — not just a result
- [ ] Both 95% and 99% (or base/adverse/severe) figures are reported as a range, not a single point
- [ ] Every assumed input (volatility, PD, LGD, haircut) is labeled with a source or "proxy" flag
- [ ] The sensitivity table shows at least 3 driver-move levels and identifies the most impactful variable
- [ ] Gross-to-net mitigation is shown explicitly for each scenario
- [ ] The summary states a specific dollar range and one recommended action tied to it
Notes
- Parametric VaR assumes normally distributed returns and understates tail risk — always report the 99% figure and note when fat tails or gap risk apply
- For credit concentration, a single large counterparty often dominates the number more than portfolio-wide PD; check for concentration before averaging
- Liquidity risk is frequently the fastest-moving of the three — model the horizon in days, not quarters, when funding rollover is involved
- When aggregating multiple risk types, state the correlation assumption explicitly; assuming independence understates combined stress-case loss
How to use this skill
This file is a Claude Code skill in Anthropic's standard format — markdown with YAML frontmatter at the top.
Installing in Claude Code:
~/.claude/skills/<folder-name>/SKILL.md— auto-loads in every session anywhere on your machine (recommended for skills you'll use across multiple projects)<your-project>/.claude/skills/<folder-name>/SKILL.md— auto-loads only when working in that project (recommended for project-specific skills)
Restart Claude Code or open a new session to register.
Using in Claude.ai, ChatGPT, Gemini, or other AI tools:
- Drag this file into a new chat, or copy and paste its contents
- Send your task or request as your next message — the AI will apply the skill where it fits
For Claude Code skill details: https://docs.anthropic.com/en/docs/claude-code/skills