Parametric cyclone cover · cat-in-a-circle

The problem

Parametric insurance pays quickly after a cyclone, but only if the trigger is well designed. One that is too tight misses damaging storms (basis risk). One that is too loose costs more than it should. Pricing it on four or five decades of history leaves a wide margin of error.

Approach

• Cat-in-a-circle trigger: payout set by a matrix of distance rings × Saffir-Simpson categories, editable for any site in the South-West Indian Ocean.
• Exact payout engine running in the browser: for each ring it solves where every track segment enters and leaves the circle, so a fast storm is never missed between two fixes.
• Historical burn analysis on IBTrACS (1980–2026) with bootstrap uncertainty.
• Three ways to extend the record: a fitted frequency × intensity model, the STORM 10,000-year catalogue (raw and with its event rate calibrated to the site) and CLIMADA probabilistic tracks, compared on frequency, intensity, closest approach and payout exceedance.
• Pricing with Wang-transform or standard-deviation loading, plus a downloadable term sheet.

Result

An interactive five-step structuring tool that runs as a static site. It also documents a known STORM bias: storms that re-intensify in the Mozambique Channel after crossing Madagascar are under-represented, so raw STORM under-prices covers there.

Next steps

• Basis risk: modelled wind at the site (CLIMADA wind fields) and modelled losses on open exposure, so each trigger design gets a basis-risk score.
• Risk layering: where this cover sits in a government’s full financing strategy (reserves, contingent credit, borrowing) is covered by the companion sovereign risk layering case study.

Limitations

Illustrative portfolio exercise on open data only, not an offer of insurance. Loadings are not market quotes.