blog · ~6 min read

Stress Testing and Scenario Analysis Construction Methods

Stress testing and scenario analysis construction covers historical replay, hypothetical shocks, and reverse stress testing with scenarios traders can run.

T By tradernewbie · Curated for beginners
#advanced#risk-management
이 문서는 영어로 되어 있습니다. 내 언어로 볼까요? Google Translate →

번역 보기에서는 대화형 도구가 작동하지 않을 수 있습니다.

Stress Testing and Scenario Analysis Construction Methods

VaR tells you about likely bad days. Stress testing tells you about unlikely catastrophic ones. The value of a stress test is not the number it produces — it is the vulnerability it reveals before the market does.

Three construction methods

1. Historical replay. Apply actual crisis returns to today's portfolio. Replay October 2008 (S&P −17% in a week, correlations to 0.9), March 2020 (volatility to 80), the 2022 bond rout (TLT −30%). Use the full vector of asset returns over the stress window, not a single index move — the cross-asset structure is the point.

2. Hypothetical shock. Specify a multi-asset shock directly: equities −15%, credit spreads +200bp, USD +8%, oil −25%, gold +12%, vol +20 points. These need not have happened; they test a coherent narrative ("flight to quality plus dollar squeeze"). Calibrate shocks to a 3-sigma to 4-sigma joint move, not arbitrary round numbers.

3. Reverse stress test. Start from the outcome and work backward: "What market move would cause a 20% portfolio loss?" Solve for the shock vector. This exposes hidden concentrations — if a single factor (e.g., equity beta) explains 80% of the catastrophic loss, the portfolio is one bet.

Building a scenario library

Maintain a fixed set of 8–12 scenarios and rerun them monthly:

  • 2008 global financial crisis (full window)
  • 2020 COVID crash (3-week version)
  • 2022 rates shock (bonds + stocks down together)
  • 2015 franc unpeg (FX gap, +30% in minutes)
  • Oil collapse (−50% in a quarter)
  • Volatility spike (VIX to 80)
  • Liquidity crisis (bid-ask widens 5×)
  • Sovereign default (single-country equity to zero, currency −40%)

Add one bespoke scenario per quarter reflecting current macro fragility.

What to measure

For each scenario, report:

  • Portfolio PnL (dollar and percent)
  • Worst single-position contribution
  • Margin call amount and available cash to meet it
  • Time to liquidate at 10% of daily volume per position
  • Funding cost if rates spike 300bp

The margin and liquidity numbers matter more than the headline PnL. A 12% mark-to-market loss is survivable; a 12% loss plus a margin call you cannot meet is not.

The honesty rules

  • Stress test the actual portfolio, not a proxy. A proxy hides the illiquid position that blows up.
  • Use stressed correlations, not normal ones. Re-running a 2008 scenario with 2024 correlations defeats the purpose.
  • Include funding and counterparty risk. A stress test that ignores prime broker failure in 2008 understated losses by 40%+.

Action triggers

Define responses before running: if a single historical scenario produces a 25% loss, cut the exposure driving it. If reverse stress test shows one factor explains 70% of catastrophic loss, diversify that factor. A stress test with no resulting action is theater.

Related market data, powered by TradingView.

Educational content · Not financial advice · Trade at your own risk