Reading a hazard score

Every location you look up in Calchis returns a Calchis Hazard Score for each peril relevant to its geography — a number from 0 to 10, with the components that produced it shown alongside. This article explains how to read one and, just as importantly, what it does not tell you.

Hazard, not risk

The score is labeled a hazard score everywhere it appears. That word is doing real work.

Hazard is the physical intensity a location is exposed to: how hard the ground shakes, how fast the wind blows, how deep the water gets. Risk is what that intensity costs you, which additionally requires knowing what is standing on the site — construction type, replacement value, occupancy, year built.

Calchis scores hazard today. Property exposure data is not yet integrated, so the platform does not claim to produce risk scores, and you will not see the word "risk" attached to a score anywhere in the interface. When exposure data is wired in, the label changes and the methodology page will say so. Until then, treat the score as the first of the three inputs an actuarial estimate needs, not as the estimate itself.

This matters when you put a number in front of a regulator or a reinsurer. A hazard score that is honestly labeled is defensible. A hazard score presented as a risk score is not.

The 0–10 scale

Scores are banded, and the interface colors them to match:

• Low — below 4.0
• Moderate — 4.0 to 6.4
• High — 6.5 to 8.4
• Extreme — 8.5 and above

Scores are per peril and are not comparable across perils. An 8.1 hurricane score and an 8.1 wildfire score do not describe the same thing, do not carry the same loss distribution, and should never be averaged into a single "hazard number" for a location. The platform deliberately does not offer such a composite.

They are also not comparable across model versions. Each score records the model version that produced it and the date it was calculated, both shown under the score bar. If you are comparing two locations, confirm they were scored by the same version.

Reading the components

Click any peril row and it expands to show the sub-scores that add up to the total. Components are drawn from three layers:

Historical frequency. How often this peril has actually affected this location in the observational record. Federal disaster declarations going back to 1953 drive the frequency sub-scores, supplemented by peril-specific catalogs — the hurricane track archive, the earthquake catalog, the federal fire perimeter record.

Hazard zone classification. Where the location sits in the authoritative federal hazard maps: FEMA flood zone, storm surge zone from the NOAA surge model, probabilistic ground motion from the USGS seismic hazard maps. These represent decades of government modeling work and are the bedrock of the score.

Physical proximity. Geometric distance to the features that drive the peril — the coastline, mapped fault traces, the wildland-urban interface.

Each component shows its own contribution and the dataset behind it. If a score looks wrong to you, the components are where you find out why, and they are the part to quote when someone asks you to justify it.

What "climate-adjusted" means

The historical record describes the climate that produced it, not the one a policy written today will be exposed to. Where the scientific literature supports a defensible adjustment for a peril and region, the scoring parameters incorporate it, and the parameter carries a description of where the adjustment came from.

Adjustments are applied per peril and per region, not as a blanket uplift. Most perils in most regions carry no adjustment at all, because the literature does not support one.

Concretely, today there are none. Until 2026-09-04 California wildfire carried a 1.4× multiplier on large-fire frequency attributed to Abatzoglou & Williams (2016); that figure was an inference from the paper rather than a number it states, so it was retired. Every peril is scored on the unadjusted historical record, which understates a warming trend. Each score still returns a climate_adjusted flag so you never have to guess which case you are looking at.

Caching and freshness

Scores are cached per location and peril, so a repeat lookup of the same address returns instantly. The generated-at timestamp under each score tells you when it was actually computed. Live event data is a separate layer and updates on its own schedule — a hazard score does not change because a storm is currently offshore. If you want to know what is happening right now at a location, that is the Active Events section, not the score.

Where to go next

The methodology page documents the scoring architecture in full, including the datasets behind each component. If you want to act on a score for a specific location, building a portfolio is the next step.