Methodology

The Fit Score

Every city gets six component scores from 0–100. Your Fit Score is the weighted average of those components using the priority sliders you set (each 0–10). Change a weight and the ranking recomputes instantly — there is no single “best city,” only the best city for your inputs.

Affordability & taxes

By default your income is treated as pre-tax. For each city we estimate your take-home pay — subtracting 2025 federal income tax, FICA, your state's real income-tax brackets, and any municipal income tax the city levies (NYC, Philadelphia, Columbus, Baltimore, Detroit, and others) — so the same salary correctly nets more in Texas or Florida than in Oregon or California, and less in NYC than elsewhere in New York. (If you enter take-home instead, flip the toggle and we use it as-is.)

Monthly cost is rent scaled to your household's likely unit size (a single person toward a 1-bedroom, larger families toward 3–4BR, using HUD bedroom-rent ratios on the 2-bedroom benchmark — HUD Fair Market Rent where available, otherwise Census median gross rent) plus a national baseline for non-housing essentials by household size: $2,450 for one person, $3,550 for two, $4,300 for three, and $5,000 for four or more — scaled by the city's real BEA price parity where imported. Where a city is still on the curated goods index, its combined state + average local sales tax is applied to the taxable share (~35%) of the baseline; real price parities are built from tax-inclusive price quotes, so they carry the tax difference already. Where transit is genuinely car-optional (EPA transit score above 50, full effect at 85+), we trim part of the baseline's transportation share — up to about 10% of the baseline. The score maps the ratio of take-home to cost: breaking even scores 50 and ~20% of headroom scores about 80.

Households with children add childcare — county median prices for center-based care, priced separately for kids under 5 and school-age kids — and get the Child Tax Credit, including its refundable portion. A single adult with children files as head of household.

Known limits: state credits and special cases (benefit recapture, millionaire surtaxes, Oregon's federal-tax subtraction) and property taxes aren't modeled; the EITC and the Child & Dependent Care Credit aren't either, so family take-home is a conservative floor. Brackets reflect 2025 law.

Coverage, goods prices & community data

Every U.S. city of 250,000+ residents is included (93 per Census 2024 estimates), plus notable mid-size cities. Everyday costs are scaled by a state cost-of-goods index (groceries, gas, household goods — Hawaii ≈113 vs Mississippi ≈90, U.S. = 100). Community political lean is the 2024 presidential margin in the city's county — shown and filtered symmetrically, as data rather than advocacy — and airport access is approximate miles to the nearest international airport with scheduled service.

A future-heat estimate projects how many days each city will reach 100°F+ around 2050 under a mid-high emissions scenario (Phoenix ≈140, Seattle ≈3) — approximate, grounded in published climate projections, and filterable so you can steer away from cities heading toward extreme heat. A companion humid-heat estimate counts projected dangerous wet-bulb days (≥77°F, the humidity-adjusted “feels deadly” threshold) — the two diverge sharply (Phoenix is dry-hot with few humid days; Houston and the Gulf Coast are the opposite), so you can steer away from desert extremes, muggy extremes, or both. Walkability (Walk Score-style, 0–100) and a natural-disaster risk composite (FEMA National Risk Index-inspired: hurricane, tornado, earthquake, wildfire, flood) round out the filters — both citywide approximations. Air quality is the annual count of days with AQI over 100 (EPA AirData; Riverside ≈95, most Eastern cities single digits), filterable, with a real EPA importer.

Filters are real filters

Region, population, income, and climate all genuinely constrain results. A city only appears as a match if your estimated take-home covers its estimated costs and, when you state a climate preference, its climate isn't a clear mismatch. If nothing passes everything, we say so plainly and show the closest options instead — ordered by the smallest shortfall — rather than quietly padding the list.

The other five components

Housing blends two percentile ranks across all cities in the dataset: median home value relative to local income (60%) and median rent relative to local income (40%) — lower ratios score higher. Jobs combines the unemployment rate (60%, lower is better) with one-year job growth (40%). Safety weights violent crime rate at 65% and property crime at 35%, both per 100k residents, lower is better — ranked against similar-sized cities (under 250k, 250k–500k, 500k–800k, 800k+), since crime rates scale with metro size and a national ranking would just reward being small. Amenities mixes the density of restaurants, arts, and recreation per 10k residents (60%, ranked against similar-sized cities) with a population-rank term (40%) that credits big-city depth. Climate is preference-relative: it starts at 100 and subtracts penalties when a city's NOAA normals miss your stated preference (for example, cold winter lows and heavy snow penalize a “warm” preference). Each dimension's penalty is capped, so one severe mismatch downgrades a city hard without erasing the ranking among otherwise-different cities.

Percentile ranks are computed within the loaded dataset, so scores are relative comparisons between the cities shown — that's also why explanations say “in this dataset.”

Data sources

Population, incomes, home values, and rents come from the Census Bureau's American Community Survey 5-year estimates. Two-bedroom Fair Market Rents come from HUD. Unemployment and job growth come from the Bureau of Labor Statistics (LAUS). Crime rates come from the FBI Crime Data Explorer. Climate normals (1991–2020) come from NOAA.

Honest caveats. Fresh installs ship with a labeled sample dataset of 72 cities using approximate figures — run the bundled importers with free API keys to load official data. FBI crime data relies on voluntary agency reporting and has coverage gaps (some large cities under-report or skip years), so treat Safety as a signal, not a verdict — and always research at the neighborhood level. City-level medians hide neighborhood variation in every metric.