How to read the Metro Guide numbers

The guide combines Census estimates, SettleSavvy neighborhood measurements and supplied metro-level data. Each answers a different question. They are not interchangeable.

Comparing metros

Each comparison names its population group. Los Angeles uses the nearest 10 other metros for its main comparisons and also shows the nearest 25, so you can see how a wider group changes the benchmark. We order metros by the absolute logarithm of their population ratio: a metro twice as large is as close as one half as large. Ties follow alphabetical metro identifiers. Population proximity does not imply matching housing markets, climates or lifestyles. Many peers of a very large metro are still much smaller.

We select the population group before checking individual measurements. Each available metro counts once in the median; missing values are excluded, with the usable count shown, and a more distant metro is not substituted. A neighborhood histogram comparison uses each peer’s neighborhood median. A metro card comparison uses the same metro measurement and units, including its stated source period and population definition. These two summaries can differ. The guide lists the selected metros and identifies any missing source dates. Similar-value popup examples are chosen by the individual metric, separately from population peers, geographic neighbors and national rankings.

Metro data and neighborhood detail

The primary dataset comes from SettleSavvy’s metro export pipeline. It combines metro-level weather, hazards and cost data with population-weighted neighborhood measurements. A neighborhood histogram shows the distribution of individual neighborhoods, so its median can differ from the population-weighted metro number. Missing observations stay missing. We do not estimate national comparisons from a partial set of guides.

SettleSavvy metro scores

These are SettleSavvy’s proprietary composite scores, ranked against metros in the same dataset. Each combines direction-adjusted factor percentiles using the weights in our macro scoring registry; available weights are renormalized when a component is missing. A higher percentile is a stronger score. A rank is relative standing in that dataset, not a guarantee about an individual address.

Affordability: home price per square foot 30%, two-bedroom rent 20%, ownership cost 10%, everyday prices 30%, gasoline 10%.
Safety: violent-crime grade 50%, property-crime grade 50%.
Climate: pleasant days 40%, summer dew point 10%, natural hazard risk 20%, disaster declarations 15%, wildfire loss rate 7.5%, riverine flood loss rate 7.5%.

Other categories combine their named member factors, including income and education for affluence, schools and playgrounds for family friendliness, and parks, trails and natural vegetation for green space. The section comparison displays the underlying measurements. Macro scores are separate from the app’s personalized neighborhood scores.

Metro honors

SettleSavvy selects up to three honors using exact factor ranks, nationally or within population classes. Large metros have at least one million residents; medium metros have at least 250,000; smaller metropolitan areas and micropolitan areas have separate classes. Large ties, repeated categories and strongly correlated measures are filtered out. An honor describes a particular measure, not an overall best-place award.

Six distinctive facts

The fact picker models the expected value at a metro’s population size, using median regression on log population. It selects unusual differences in either direction, filters correlated and semantically similar facts, and avoids repeating honors. Climate and hazards have dedicated sections. A demographic or religious difference is descriptive; it is not a recommendation about who should live in a neighborhood.

Similar overall score, different strengths

Named section examples use a different selection. Candidates must have at least 150,000 residents and at least 15% of the subject metro’s population. We start within five percentile points of its section score, widening to nine and then fourteen until at least eight candidates qualify, or the widest range is reached. We examine at most the 80 closest section percentiles.

At least one component percentile must differ by 22 points, reduced to 12 when the subject score is in the top 5%, and eight in the top 2%. Candidates are ranked by the weighted absolute differences in their direction-adjusted component percentiles. The selection adds 0.30 when each metro leads on a component by at least 12 points, and 0.15 for a metro in a fixed familiar-city list. Tied totals retain the prior order by section-score proximity. The guide shows the selected examples, scores and reasons a candidate was omitted. A similar composite percentile can combine very different housing prices or other measurements. A metro can belong to the population group without being chosen as a named section example.

Housing, income and everyday prices

Census rent includes rental homes of different sizes and utility arrangements. Home value is what owners estimate their property is worth. These are metro estimates. Two-bedroom rent, ownership costs and modeled price per square foot are separate neighborhood measures; they do not describe the same set of households or homes.

SettleSavvy’s modeled home prices are estimates, not property appraisals or listing quotes. The stored factor definition describes adjustment using the FHFA house-price index. The everyday-price index excludes housing and uses a US average of 100.

SettleSavvy crime grades

Safety grades use SettleSavvy’s proprietary crime model, not a raw count of police reports. The published methodology describes county crime totals, modeled neighborhood allocation and a denominator that includes residents and workers. Higher letter grades mean greater modeled safety. A grade cannot predict whether an individual will experience a crime.

Read the full crime grading methodology and validation.

School grades and access measures

School grades summarize district test performance using the app’s letter-grade conversion. They do not evaluate every aspect of a school. When coverage is partial, the guide identifies that in the metric’s helper text. Places counted within a radius show availability, not travel time, service frequency or a guaranteed walking route.

Climate

Weather figures come from the supplied metro export, which combines county climate normals using population weights. They describe a typical climate rather than a forecast. “Pleasant days” means a normal high between 65°F and 82°F and a low above 45°F. Weather and conditions at an individual address can differ.

Suggested places

Place suggestions use the highest available SettleSavvy factor scores among places in the metro. These are proprietary composite scores. Winning on one factor does not make a place the best overall, and suggestions are not matched to demographic groups. The current guide does not publish a full formula or validation study for every composite factor.

Data sources

Savvy Rankings and the metro guides are built from these datasets. Each guide also cites the source and period behind every number it shows.

Source records and research

Source definitions, observation periods, export fingerprints and evidence checks remain attached to the saved backend version. Web context links to primary sources and is kept separate from numerical measurements. A successful URL check confirms reachability, not the truth of every claim. Sources and charts update when a new guide version is prepared and published, not on every page view.