Core Path · L6.2.02

Risk factor mapping and proxies

14 minRisk factorsCash-flow mappingProxiesTracking errorNegative rates

By the end you will be able to

  1. 1Choose risk factors for a curve: zero against par rates, the vertices, and absolute against relative changes when rates go negative.
  2. 2Map cash flows onto vertices so that present value and PV01 (or variance) are kept, and compare the maps' VaRs.
  3. 3Fit a proxy for a factor with no usable history, and read its R², tracking error and stability.
  4. 4Measure what a poor proxy costs in VaR and in backtests, and how a short history is filled and screened.

Why it matters

In September 2022 the Bank of England's 2-year OIS rate rose 148bp in a month. The VaR engine of L6.2.01: Anatomy of a VaR engine has no sterling curve history of its own, so it moves sterling rates with a proxy built from US and euro 2-year rates. For that month the proxy said 71bp. On the sleeve's sterling OIS alone, the gap is worth about £2.3m. An engine sees only the risks its factors can move, and only as well as their histories let it.