Every figure on this site is derived from published federal data by the arithmetic below. Where the model makes a judgement — service life, ticket price, storm weighting — the number is stated here and can be replaced with your own in any market file.
Housing stock as a vehicle fleet: some units have the part, it lasts a fixed number of years, and the fleet runs on smooth roads or on gravel.
Addressable homes are capped at single-family detached stock, not owner-occupied units — 500 condo owners do not buy 500 roofs. That single constraint is the difference between New York appearing to be the country's largest roofing market and appearing as what it is.
Those homes are divided by the line's service life to give a replacement rate, then adjusted by an age multiplier where the stock is older than the national median and a storm multiplier where the hail record justifies it. New construction is added at a discounted ticket, because builder work does not price like retail replacement.
| Service line | Service life | Eligible stock | Ticket | Storm sensitivity |
|---|---|---|---|---|
| Roofing | 22 yrs | 82% | $17.60K | 0.85 |
| Windows & Doors | 25 yrs | 88% | $11K | 0.30 |
| Siding & Exterior | 30 yrs | 70% | $16.50K | 0.60 |
| Gutters & Trim | 24 yrs | 85% | $2.40K | 0.55 |
Roofing's ticket is anchored to Verisk's 2025 measured average of $17,631. The other three are industry rules of thumb, and are the assumptions most worth replacing with your own.
Seven components, every weight published. The composite is a sort key, not a verdict.
| Component | What it measures | Weight |
|---|---|---|
| Market size | Absolute dollars of work available | 28% |
| Growth | Permit intensity and construction momentum | 17% |
| Ability to pay | Household income and home values | 14% |
| Storm exposure | Damaging hail, per 1,000 sq mi, last 10 years | 11% |
| Climate load | Heating and cooling degree days | 11% |
| Open market | Homes per specialty-trade contractor | 10% |
| Housing age | Median age of the housing stock | 9% |
Scores top out at 82.9 and the median is 50.8, so a market above 70 sits in the top 5% of the country. Every score is published with its national rank for that reason — 82 out of 100 reads like a B, when in fact it is the strongest roofing market in the United States.
Stated in full. A limitation found here is one nobody has to discover in a committee.
Seven components — market size, growth, ability to pay, storm exposure, climate load, open market and housing age — are each scored as a percentile against all 935 US metros, then combined as a weighted average. The weights are published in full above and total 100%.
Repair-and-remodel figures are heavily right-skewed. On a z-score the largest one or two metros compress everything beneath them toward zero, and the ranking stops discriminating among the several hundred markets where most expansion decisions are actually made.
Public federal sources only: the US Census Bureau's American Community Survey and Building Permits Survey, the Bureau of Labor Statistics' Quarterly Census of Employment and Wages, and NOAA Storm Events and Climate Normals. No licensed panel data and no vendor model.
Contractor counts resolve for 393 of 935 metros, so the rest are scored on six components rather than seven. Coverage is whole-metro rather than drive-time. Replacement is treated as steady-state, which understates the wave pattern in fast-built markets. Insurance dynamics are not modeled, and scope is residential only.
No. They are planning estimates built from published data and documented assumptions, designed to rank markets against one another on a consistent basis. That is a different task from predicting what a specific branch will bill, and the model does not attempt the second.