AI for Market Risk Management: VaR, Stress Testing, and FRTB with Claude (2026)

Market Risk Management and AI

Market risk teams spend enormous time on VaR production, backtesting, P&L attribution, and regulatory reporting. Each requires combining risk model output with financial data in ways that are analytically intensive but structurally repetitive. Claude with ClaudeFinLab accelerates the analysis, interpretation, and documentation work — explaining model behavior, identifying risk concentrations, drafting risk committee presentations, and navigating FRTB requirements.

VaR Calculation and Interpretation

• "Compute 10-day 99% VaR for a fixed income portfolio using historical simulation: portfolio has 3 positions — 5-year UST $50M, 10-year UST $30M, 30-year UST $20M. Use 500 days of historical rate moves. For each historical scenario, apply the rate change to each position's duration to compute daily P&L. The 5th worst scenario (99th percentile of 500 days, 10-day scaling: multiply 1-day VaR by √10) gives the VaR estimate. Current DV01: 5Y $24,000, 10Y $28,000, 30Y $38,000."
• "Decompose VaR by risk factor: total portfolio 10-day 99% VaR $2.8M. Risk factor attribution: interest rate risk $2.4M (DV01 exposure to parallel shifts), yield curve risk $0.6M (non-parallel moves), credit spread risk $0.3M (MBS OAS widening), FX risk $0.2M (EUR exposure). Diversification benefit: -$0.7M (correlations reduce total below sum of parts). Identify the largest single risk factor and the position driving it."

Stress Testing and Scenario Analysis

• "Run the following stress scenarios on the portfolio: (1) 2008 Global Financial Crisis: rate curve steepening +150bps long end, credit spreads +400bps, equities -40%; (2) 2020 COVID shock: rates -100bps, credit spreads +250bps, equities -35%; (3) 2022 Rate Shock: parallel rate +350bps, equities -20%; (4) Hypothetical tail event: rates +500bps, credit spreads +800bps. For each scenario, compute total portfolio P&L. Identify which scenario is most severe for our current positioning."
• "Design a reverse stress test: instead of applying scenarios and computing losses, work backward. What rate move would cause a loss of $15M (our Board-approved risk limit)? At current DV01 $90,000/bp, loss of $15M requires a rate move of 167bps (parallel shift). Is that plausible? What is the historical probability of a 167bps parallel shift in a 1-month window? Assess whether the Board limit is calibrated appropriately to current market risk."

Options Greeks and Risk Management

• "Compute the Greeks for this options portfolio: [list of 8 positions with underlying, strike, expiry, notional]. Delta: sum of position deltas × current underlying prices = total portfolio delta exposure in $. Gamma: rate of change of delta per $1 move in underlying. Vega: sensitivity to 1% implied vol move. Theta: daily time decay. Rho: sensitivity to 1% interest rate change. Identify which position dominates each Greek. What is the net delta and what hedge trade (buy/sell underlying) would make the portfolio delta-neutral?"
• "Assess vega concentration risk: portfolio is long $2.8M vega primarily from 3-month ATM options on S&P 500 and EUR/USD. VIX is currently at 14 (near historical lows). Risk: if implied volatility spikes (VIX 14→28, as in October 2023 or September 2022), the portfolio would gain $2.8M × 14pp = $39.2M. But if vol remains compressed, daily theta costs are $185,000/day (=vega/market conventions). Assess the asymmetry and whether the position is appropriately sized."

VaR Backtesting

• "Run the VaR backtest for the quarterly regulatory submission: 99% 1-day VaR for 250 trading days. Actual P&L was worse than VaR on 4 days (exceptions). Basel traffic light: Zone Green (0-4 exceptions) → our 4 exceptions are in Green Zone (multiplier 3.0). Zone Yellow (5-9 exceptions) → would trigger multiplier increase to 3.4-3.8. Zone Red (10+ exceptions) → model review required. Document the 4 exceptions: dates, P&L, VaR, and likely causes (model limitations vs genuine tail events)."

FRTB Standardized Approach

• "Compute the FRTB SA capital charge for a corporate bond portfolio: Delta risk charge = sensitivity × risk weight. For an IG corporate bond bucket (Bucket 3, BBB, 3-5 year tenor): risk weight 1.0%. Net sensitivity $42M. Delta risk charge: $420K. Curvature risk charge: 0.5% of $42M = $210K. Residual risk add-on: 0.1% for non-vanilla instruments. Aggregate across all buckets using FRTB correlation matrix. Total SA market risk capital requirement."

Where to Start

Connect ClaudeFinLab's portfolio risk MCP server — it computes VaR, EVT tail risk, and stress scenarios directly from position data. For FRTB and regulatory capital, describe your portfolio composition and ask Claude to walk through the SA risk weight methodology. For backtesting, paste the daily VaR and P&L time series and ask Claude to compute the exception count, classify the Basel zone, and draft the regulatory documentation.