How the Roof Stress Index is built
The index asks one question of each county: how much of the housing stock is old, how often does severe hail arrive, and how slowly is the stock being replaced? It uses only files that anyone can download. It does not use our leads, prices or any opinion.
The three inputs
- Old homes. Share of housing units built before 2000, from the Census Bureau's American Community Survey 5-year estimates. Median county: 80.4%.
- Severe hail days. Average number of days per year on which NOAA's Storm Events Database holds at least one hail report of 1 inch or larger in the county, counting full calendar years only (2014–2026). Median county: 2.0 days. NOAA Storm Events.
- New-home permits. Single-family permits per 1,000 housing units, 2014 to 2025, from the Census Building Permits Survey. More permits means the stock is renewed faster, so this input counts against the score. Median county: 35.5.
The calculation
Each input becomes a mid-rank percentile among the 1,113 scored counties (0 for the lowest value, 100 for the highest; ties share the middle). The permit percentile is flipped (100 minus the percentile). The score is the plain average of the three. Rank 1 is the highest score. The band shows position, not a score cutoff: the top 10% of counties are "very high", the next 20% "high", the next 40% "moderate" and the lowest 30% "lower".
A state percentile on a county page uses the same recipe among that state's scored counties and is shown only when a state has at least 8.
Who is left out
We scored counties with at least 20 hail reports in the NOAA file, at least 1,500 housing units, and both Census files present: 1,113 of the 1,275 counties in our hail dataset. Below those limits a ratio can swing on a few dozen homes. A county with few hail reports is not hail-free: NOAA data depends on people filing reports, and thinly populated areas file fewer.
Does the weighting change the answer?
We re-scored every county with three other weightings and counted how many of the top 50 counties (equal weights) remain among the top 100 and top 50.
| Weighting (age : hail : permits) | Still in top 100 | Still in top 50 | County ranked first |
|---|---|---|---|
| equal weights (1:1:1) | 50 of 50 | 50 of 50 | Sherman County, KS |
| age counts double (2:1:1) | 50 of 50 | 42 of 50 | Sherman County, KS |
| hail counts double (1:2:1) | 50 of 50 | 40 of 50 | Sherman County, KS |
| permits count double (1:1:2) | 50 of 50 | 44 of 50 | Hutchinson County, TX |
What it cannot tell you
- It describes a county, not your roof. Two houses on one street can differ by decades.
- Old housing does not mean old roofs: many homes have been re-roofed. The index measures exposure, not condition.
- A hail day counts when at least one report of 1 inch or more exists anywhere in the county. Large counties collect more reports by size alone.
- Permit counts cover single-family homes only and come from places that issue permits and report them to the Census Bureau.
- Nothing here is a forecast, an appraisal or an insurance opinion.
Questions people ask about the score
Does a high score mean my roof is damaged?
No. The score describes the county, not your house. It says that older homes, frequent severe hail and slow renewal tend to go together there, so roofs spend more of their life exposed. Whether yours is damaged needs a look from a roofer or an adjuster.
How much of the score is hail and how much is age?
They are separate inputs with equal weight. Each county page shows which of the three pulls hardest, and the table above shows how little the top of the national list moves when one input counts double.
How were the roof-life ranges in the calculator chosen?
They are general planning figures for each material, not a promise. A warranty, the climate and an inspection decide the real answer, and hail days in the calculator show exposure of the county, not damage to a given roof.
Why is my county missing?
Either it is below the report and housing limits listed under "Who is left out", or its page has not been released yet. Pages roll out in weekly batches and a county page stays hidden until its release.
Sources
NOAA Storm Events Database for hail reports (2014–2026), Census ACS 5-year estimates for housing age, and the Census Building Permits Survey for single-family permits (2014 to 2025). The index is built from public data and is not an inspection, an appraisal or a forecast.
Data as of 2026-10-06. Back to the county map, or check your own address with the hail check.