Three Cat-in-a-Box Hurricane Triggers, Replayed Against 36 Years of Best Track

A cat-in-a-box trigger pays when a hurricane's center passes through a defined box at or beyond a stated intensity. The terms are short and the answer is objective, which is why the structure is common in parametric hurricane cover. What the terms do not tell you is how often they would have paid. The historical record can.

Calchis replays determination/2 terms year by year against NHC's best track. The replay is not a second calculation: each season is settled by the same evaluator that settles a live trigger, with the risk period moved to that year. We wrote one illustrative schedule and placed it over three coastlines. It is not the term sheet of any bond.

The terms

• Rule: the lowest central pressure while the storm's center is inside the box, read at NHC's published best-track fixes.
• Tiers: 25% of principal at or below 965 mb, 60% at or below 950 mb, 100% at or below 930 mb. The greatest tier reached in a season pays, once.
• Risk period: June 1 to December 1.
• Boxes: South Florida 25.0–26.5°N, 81.0–79.8°W; Tampa Bay 27.0–28.5°N, 83.0–82.0°W; Houston-Galveston 28.8–30.0°N, 95.8–94.3°W.
• Record: 1990 to 2025, 36 seasons. Pressure is the reason the window opens in 1990: under 7% of North Atlantic best-track points carry a pressure before the 1950s and 78% in the 1980s, against 98% in the 1990s. A season with no pressure on record would read as a season that did not pay.

What would have paid

Box	Season	Storm	Deciding fix	Payout
South Florida	1992	Andrew	922 mb, 25.5°N 80.3°W, 09:05 UTC Aug 24	100%
South Florida	2005	Wilma	950 mb, 26.2°N 81.0°W, 12:00 UTC Oct 24	60%
Houston-Galveston	2008	Ike	950 mb, 29.3°N 94.7°W, 07:00 UTC Sep 13	60%
Tampa Bay	2017	Irma	961 mb, 28.2°N 82.2°W, 06:00 UTC Sep 11	25%
Tampa Bay	2024	Milton	956 mb, 27.3°N 82.7°W, 00:00 UTC Oct 10	25%

Box	Paying seasons	Expected annual payout	Storms in the box that paid nothing
South Florida	2 of 36	4.4% of principal	7
Tampa Bay	2 of 36	1.4%	7
Houston-Galveston	1 of 36	1.7%	7

Two or fewer paying seasons is an order of magnitude, not an estimate, and the replay says so on its own result. It is a reading of what happened under today's terms, with no climate trend and no change in exposure.

Two payouts sit exactly on a line

Wilma's deciding fix is at 81.0°W, the South Florida box's western edge. Ike's landfall record reads 950 mb, the 60% threshold exactly. Both pay because the terms say the edge and the threshold count. Move either by a tenth of a degree or a millibar and the 2005 or 2008 season pays less or nothing. Whether a boundary is inclusive is not a detail a calculation agent should have to infer.

Two storms the boxes miss

Harvey, 2017. Harvey's losses in Houston came from rain. Its center came ashore near 27.8°N 96.8°W at 941 mb, southwest of the Houston-Galveston box. Its fix nearest the box, at 18:00 UTC on August 29, was 28.5°N 94.2°W at 993 mb, just to its southeast and far above every tier. The box pays nothing for 2017.

Irma, 2017. Irma's 18:00 UTC fix on September 10 was at 25.6°N 81.7°W at 936 mb, on Florida's southwest coast, 0.7° west of the South Florida box. The box pays nothing for it. Twelve hours later Irma pays 25% in Tampa Bay, at 961 mb, after it had weakened.

Neither is a flaw in the evaluator. They are the basis risk the structure carries, visible in advance when the terms are run against the record.

What the replay is not

A modeled probability. Thirty-six seasons cannot give a 1-in-100 figure. A replay tells you which storms the terms would have caught and missed, not the price.

Fixes only, by these terms. The center is tested at NHC's published fixes. determination/2 can also test the straight line between fixes, and the terms must say which. These do not. IBTrACS adds its own positions every three hours between NHC's fixes; the replay sets those aside, because they are not NHC records.

Best track, not the advisory. A live trigger may settle on the operational advisory. The record of past seasons is the best track, which NHC revises after the season.

Sources: NOAA National Hurricane Center HURDAT2 best track, via the IBTrACS North Atlantic record (NOAA NCEI), seasons 1990–2025; 2025 is provisional until NHC finalizes it. The terms and the evaluator are the determination standard; any parametric trigger in Calchis can be backtested the same way.