Market Risk
What the exam tests
This domain tests market risk types, sensitivity measures, VaR and expected shortfall, stress testing and limit management. Questions often ask you to interpret a VaR number or choose the right sensitivity for a risk.
Core rules and concepts
- Market risk is loss from adverse movements in market prices: interest rates, equities, FX, commodities and credit spreads.
- Interest rate risk includes repricing, yield curve, basis and option risk. Duration measures sensitivity to parallel yield changes; DV01 or PV01 is the change in value for a one basis point move.
- Equity risk includes delta (price sensitivity), gamma (change in delta), vega (volatility sensitivity) and beta (sensitivity to the market).
- FX risk arises from currency positions and translation exposure. Commodity risk includes outright price, basis and shape risk.
- VaR estimates the maximum loss over a horizon at a confidence level. A 1-day 99% VaR of £2m means a 1% probability of a loss greater than £2m, not a maximum loss.
- Expected shortfall (ES) is the average loss in the tail beyond VaR. Basel's FRTB moves from VaR to ES because VaR ignores tail severity.
- Stress testing and scenario analysis test large, plausible shocks that VaR may miss. Back-testing compares VaR exceptions with expected exceptions; too many exceptions indicates model failure.
- Limits include notional, sensitivity, VaR and stop-loss limits. Hedging with futures, forwards, swaps and options reduces risk but introduces basis, correlation and liquidity risk.
Worked example
A portfolio has a 1-day 99% VaR of £1.5m. A manager scales it to 10 days using the square-root-of-time rule: £1.5m x the square root of 10, about £4.74m. This assumes independent, identically distributed normal returns and no autocorrelation. If the portfolio is illiquid or returns are serially correlated, the 10-day VaR is understated. In back-testing 250 trading days at 99%, the expected number of exceptions is 2.5. Six exceptions would be a warning that the model or volatility estimate is wrong; the firm should investigate, not simply raise the limit.
Traps that cost marks
- Calling VaR the worst-case loss. It is a quantile, not a maximum.
- Mixing confidence levels: 95% VaR is smaller than 99% VaR for the same portfolio.
- Scaling VaR by the square root of time without checking assumptions.
- Using duration for a non-parallel curve shift or for a portfolio with options.
- Ignoring liquidity and basis risk when hedging.
- Relying on VaR alone and omitting stress testing and expected shortfall.