Methodology

Every number, explained

HomeResilience Labs produces simplified projections based on public data and established relationships — never "scientific simulation," never advice, never a quote. This page is the full recipe: sources, formulas, assumptions, and what we deliberately don't claim.

The disclaimers, up front. Nothing here is financial, insurance, or engineering advice. Insurance figures are illustrative ranges based on public program descriptions — your carrier decides your discount, not us. Costs are national ranges; local quotes will differ. Consult qualified professionals before structural work.

01 Data sources

Source	What we use	License / access
FEMA National Risk Index (v1.20, Dec 2025)	Per-county risk scores (0–100), ratings, expected annual losses (EAL), building-loss rates (ALRB), event frequencies for 18 natural hazards; social vulnerability and community resilience scores.	U.S. Government public domain; fetched from FEMA's ArcGIS feature service.
Census 2020 ZCTA–county relationship file	ZIP → county resolution. Where a ZIP spans counties, the county with the largest land-area overlap wins.	Public domain.
U.S. EIA	Average residential electricity price by state (¢/kWh), for the energy-savings stream.	Public data; refreshed from the EIA API.
Curated mitigation-effectiveness catalog	Per-retrofit risk-reduction fractions, installed-cost ranges, useful life, energy savings, and illustrative insurance-credit ranges — aggregated from FEMA P-804/P-312/P-530, NIBS Mitigation Saves, IBHS, DOE/ENERGY STAR, and state program documents. Each retrofit lists its citations in the app.	Curated by us; sources linked per retrofit.
Future-adjustment multipliers	Directional mid-century (~2050) multipliers by census region and hazard, informed by NCA5 / CMRA direction-of-change statements. Always ≥ 1 (we never claim a hazard will improve). Shown with their basis wherever used; the future toggle defaults off.	Curated; basis string shown in-app.

02 The resilience score

The score is anchored to your home's total expected annual loss rate — dollars of expected building loss per dollar of home value per year, summed across all 18 hazards:

rateₕ = ALRBₕ × vₕ × fₕ     ALRBₕ = FEMA building-loss rate for hazard h
                             vₕ = your home's vulnerability factor
                             fₕ = future multiplier (1 unless toggle on)
ALR   = Σₕ rateₕ
Score = 100 × e^(−ALR / 0.35%)

Anchors: a home with no expected losses scores 100; a typical U.S. loss rate (~0.1%/yr) scores about 75; a hard-hit coastal profile (0.35%/yr) scores about 37; extreme profiles (≥1%/yr) fall to single digits. We deliberately do not score against FEMA's 0–100 risk scores — those are national percentile rankings, so nearly every populous county sits near the top and scoring against them would tell almost everyone their home is doomed. Loss rates are the honest signal; percentiles appear in the hazard bars for context only.

Vulnerability factors v_h start at 1.0 and are nudged by your inputs — for example, a 20-year-old roof adds +0.25 to wind-hazard vulnerability; a basement adds +0.25 to flood; pre-1980 construction adds +0.15 to thermal hazards; pre-1990 adds +0.20 to earthquake. All factors are clamped to [0.6, 1.6]. Higher score = more resilient.

03 Retrofit ROI

Each retrofit r has a per-hazard loss-reduction fraction δ_h,r from the curated catalog. Three annual benefit streams:

Avoided loss   Aᵣ = Σₕ ALRBₕ × homeValue × vₕ × fₕ × δₕᵣ
Energy         Gᵣ = kWh_savedᵣ × your state's ¢/kWh (EIA)
Insurance      Iᵣ = annualPremium × [discount range]   ← illustrative only

Benefit        Bᵣ = Aᵣ + Gᵣ + mid(Iᵣ)
Payback        = installed cost (midpoint) / Bᵣ
NPV            = Bᵣ × annuity(life, 4%) − cost
Resilience ROI = NPV / cost                    ← the ranking key

ALRB is FEMA's annualized loss rate for buildings — county building losses per dollar of building value — which we scale by your home value and vulnerability. This keeps avoided-loss dollars anchored to observed federal loss data rather than invented damage curves. Where a hazard's building-loss rate is unavailable (e.g. drought, which FEMA models as agricultural loss), the retrofit earns nothing from that hazard.

04 Bundles & the budget optimizer

When several retrofits reduce the same hazard, reductions apply multiplicatively — two measures with δ = 0.5 and 0.4 leave 0.5 × 0.6 = 30% of the loss, avoiding 70% (not a double-counted 90%). The "best bundle under $X" search runs a 0/1 knapsack (dynamic programming on a $50 cost grid) plus a marginal-benefit greedy that re-computes interactions after each pick; both candidate bundles are valued with the exact multiplicative math and the better one is shown. Stacked insurance discounts are capped at 35% of premium as a guardrail. Bundle NPV discounts the combined stream over a cost-weighted mean useful life.

05 What we deliberately don't claim

No parcel-level precision. Hazard data is county-level. Your side of the creek matters and we can't see it.
No insurance promises. Discount ranges summarize public program descriptions (e.g., FORTIFIED credits, CEA brace-and-bolt, wind-mitigation forms). Your premium is between you and your carrier.
No future-certainty. The ~2050 toggle applies published-direction multipliers, clearly labeled, default off.
No engineering judgment. Effectiveness fractions are literature midpoints for typical homes, not an inspection of yours.

06 Privacy, reproducibility & funding

Privacy: the ranking math runs in your browser. Your home details reach our servers only if you request a PDF report or save a plan — and then only those inputs, no name or account.
Reproducibility: every data import writes a dated manifest (source, version, row counts, hash). Results are labeled with the FEMA NRI version they used.
Funding disclosure (FTC): retrofit detail pages include outbound links, some of which may earn a commission at no cost to you. Rankings come from the formulas above and are never influenced by partners. Sponsored placement is not sold.

07 Primary references

Run the simulator →