Insights

In March we wrote that most cat bonds use parametric triggers. They do not: of the 245 outstanding bonds we track, 7 do. The correction is the point. The modeling agent works before the event and the calculation agent after it; a fund manager holding a parametric position decides in the hours between, from broker emails. What a public federal feed can and cannot say in that window.

PCS says average annual U.S. severe thunderstorm losses rose from about $4 billion to over $40 billion, and Verisk is breaking its industry loss index down to county level so county-scale bonds can be priced. A county emergency manager's jurisdiction and an index bond's coverage are the same polygon. What we hold at that resolution today, from the federal record, and what we do not.

Every hospital, nursing home and senior-living operator keeps a hazard vulnerability analysis for the surveyor. The probability column is usually a number typed from memory; the county's federal record already holds the answer. How we pre-fill one column, leave the other six alone, freeze the ratings when they are signed off, and turn the hazards you rated high into the conditions the building is watched for.

Twelve Securis expects capital inflows and a quiet catastrophe year to push January 1 rates down, and says the job for managers is to tell normal repricing apart from a deterioration in the risk itself. The distinction is a data problem. Repricing is one number everyone can see. Deterioration hides in attachment points, covered perils and subject-business definitions that almost nobody reads after the deal closes. Of the 245 active cat bonds we track, we hold a published attachment point for seven.

Swiss Re Institute puts the premium opportunity from AI data centers and renewable energy at $200 billion between 2026 and 2030, and sees a role for cat bonds and sidecars in the upper layers of the towers. The catch is where the assets are. Data center exposure is the least diversified property book in the world, solar and wind farms sit in the hail and convective belts, and the capital above the tower will need to price the hazard site by site.

Every vendor selling an EOC platform says it auto-generates ICS forms. The Incident Status Summary runs to 42 blocks, and most of them ask for a judgment rather than a value. Here is which blocks a system can fill from data, which it cannot, and why the difference matters when you file one.

The NIFC WFIGS current perimeter service returns about 200 features. It also returns 36 MB and takes nearly 40 seconds. Most ingest code assumes those two facts cannot both be true, and quietly goes stale when they are.

The insurance-linked securities market has grown to $61 billion in outstanding cat bonds, yet fund managers still monitor parametric triggers using spreadsheets and email alerts from broker desks. The gap between data availability and tooling is a market failure waiting to be solved.

Every Calchis hazard score traces to a peer-reviewed publication or government technical report. This post explains the scoring architecture, the data sources that feed it, and why we show the math instead of hiding it behind a proprietary label.

Hurricane Harvey dumped 60 inches of rain on Houston in 2017, causing $125 billion in damage. Single-peril models that evaluated wind risk and flood risk independently missed the compound amplification that made Harvey catastrophic. Here is what the data shows.

Catastrophe models price a bond before the event and a calculation agent settles it after. A parametric trigger needs a physical measurement in between. Here's why that gap matters for the $61B ILS market.

The three-module architecture behind every catastrophe model — hazard, vulnerability, financial — and why transparency in 2026 beats the black boxes built in the 1990s.

Analysis of 40 years of NIFC fire perimeter data reveals accelerating large fire frequency in California and the Pacific Northwest. Here's what it means for property underwriting.

With Atlantic sea surface temperatures running well above average and a La Nina-to-neutral ENSO transition underway, the 2026 hurricane season outlook carries significant implications for the $61B cat bond market. Here is what the early forecast signals mean for parametric trigger exposure.

FEMA flood zone designations determine insurance requirements, property values, and mortgage eligibility for millions of US properties. Here is what each zone means, how they are determined, and why they matter more than most homeowners realize.

The USGS National Seismic Hazard Model identifies significant earthquake hazard in 42 US states. California gets the attention, but the data reveals concerning probabilities in the Pacific Northwest, Central US, and South Carolina.

Catastrophe bonds are a $61 billion asset class that most finance professionals have never encountered. Here is a plain-English explanation of how they work, why they exist, and why the market is growing at 15% annually.

The OpenFEMA endpoint most flood models read from carries a deprecation date, and its replacement is renamed rather than re-versioned — so the obvious check for a successor returns 404 and reads as “none exists”. Three things change at once, and one of them fails silently.

Parametric basis risk is usually discussed as a structuring trade-off — tighter triggers, more boxes, more layers. Much of it is upstream of structuring: it is measurement latency, spatial resolution, and revision policy in the underlying feed.

A parametric trigger evaluation that stores only a pass/fail is not an audit trail. Reconstructing a decision months later requires the source, the timestamp, the value, the threshold, and the distinction between not-breached and not-evaluated.