Methodology

Built on data, not opinion.

Every score is a national percentile. Every signal has a published definition. Every location carries a confidence tier. When data is sparse, we tell you — we never hide it behind an inflated score.

You rank the signals

The Demand Score

Each location receives a single Demand Score between 0 and 100. A score of 87 means that location is in the top 13% nationally, on the signals you ranked, in the most recent month. Higher means tighter supply and stronger rental demand as you have defined it.

The score is calculated by re-normalized exponential decay across the signals you ranked. Your top-ranked signal carries the most weight; each subsequent rank carries 60% of the rank above it. Signals you toggle off contribute zero. The remaining weights are renormalized to sum to 1.0 — so the score is always comparable across users with different rankings, even though each person's ranking is their own.

weight(rank n) = 0.6^(n - 1)
score = Σ ( normalized_signal × weight ) / Σ weights × 100

Each signal is converted to a 0–100 percentile using all valid US locations in the current month. That means every score is grounded in a national context — not a local one. A 90 in Spokane and a 90 in Austin are directly comparable.

What we measure

The 11 leading signals

The signals were chosen because they lead price, not lag it. They reflect the buying, listing, and renting decisions that move markets — captured at the location level for every US location with sufficient data.

What to buy, not just where

Property Standouts

The Demand Score tells you where demand is strongest. Property Standouts goes one level deeper — what to look at within a location. When you expand any ZIP, two cards surface the standout asset type for two different goals:

An "asset type" is a property type × bedroom countsegment — for example, a 3-bedroom single-family home, or a 1-bedroom condo. Each card shows that segment's median price, typical rent, dwelling size, typical age, lot size, days on market, sale-to-list ratio, price-cut rate, and its headline yield or appreciation.

How the figures are derived:

• Source: medians pooled from roughly the last 18 months of recorded home sales in each location.
• Appreciation: the annualized growth between the two most recent sale prices of the same property (a repeat-sales method, the principle behind the Case-Shiller index), taken as a median across qualifying homes — so it reflects real buyer-to-buyer price changes, not a model estimate.
• Minimum-sample gating: a segment must have enough recent sales to qualify, and appreciation requires enough repeat-sale pairs. A thin, lucky cell can never win — if the data is too sparse, the cards simply do not appear for that ZIP.
• Coverage honesty: Property Standouts are shown only where the sales record supports them. They are naturally absent in non-disclosure states and very thin rural markets — we would rather show nothing than something unreliable.

Property Standouts describe past, recorded sales for information only. They are not a recommendation to buy or sell any property, and not investment advice — see the note at the bottom of this page.

Data quality

Confidence tiers

Each location carries a confidence tier reflecting two things: how much underlying data sits behind its score, and how stableits signals are month to month. A location with full data coverage but signals that swing wildly between months (for example, days-on-market jumping from 30 to 105) is downgraded — volatile inputs make a single month's score less dependable, and we would rather say so than present false certainty. Locations with too little signal coverage are excluded entirely — we will not invent a score where one cannot be supported.

By default the app shows Very High and High tiers only. You can expand the filter to include Medium or Low — but we recommend treating those scores with corresponding caution.

Where the data comes from

Our data principles

Raw data is aggregated from a mix of free public sources and paid commercial feeds, then transformed into the leading indicators above. We do not rely on a single source for any signal — every metric is cross-checked or derived from at least two independent inputs where possible.

• Transparency: The full formula above is the entire formula. There are no hidden adjustments, no editorial overrides, no "trust us" layers.
• Coverage honesty: Locations with insufficient data are excluded from results. Their absence is reported in the footer along with the eligible count.
• Comparability: Every signal is normalized against the full US universe of locations in the current month. Local rankings are derivative of national context, not independent of it.
• Reproducibility: Given the same inputs, the same ranking produces the same scores every time. No randomness, no machine-learning black boxes.

Cadence

Updates and history

Scores update monthly. Pro users can view previous months alongside the current month using the same column layout, so they are directly comparable. The number of historical months available may grow over time.

The trajectory sentence shown when you expand a location explains the largest drivers of month-over-month score change. If a signal was missing in the previous month for more than 95% of locations (a common occurrence for newly added signals), the trajectory and delta are suppressed entirely to avoid misleading comparisons.

Important

Abunsh is a data and analytics tool. The Demand Score is a ranking, not a recommendation. Past performance is not indicative of future results. Abunsh does not provide investment, legal, tax, or financial advice. Always do your own due diligence and consult appropriate licensed advisors before making real-estate investment decisions.