The principle: every number must survive being questioned
Most "market data" online is a Zestimate, a screenshot, or a confident decimal built on a sample too thin to trust. We hold our numbers to the standard we use to underwrite our own purchases: each figure traces back to a real source, and we show the sample size so you can tell a strong number from a weak one. If a number cannot survive being questioned, it should not drive a six-figure decision.
Where the data comes from
Parcels, ownership, and assessments come from Shelby County records. Sales and deed history come from recorded transactions. Permits and code activity come from the county and city. Long-term rent comps come from live and recently leased listings across the major property-management portals and rental sites in the metro, refreshed nightly. Short-term figures come from real booking comps over the trailing twelve months. None of it is an LLM guess or a stale automated estimate.
All of it is joined on a single spine keyed to the parcel, so when you type an address we are reading that property, not a ZIP-code average standing in for it.
How a rent estimate is built
We match your address to its parcel, read its bedroom count and size, then pull recent rent comps for the same bedroom count in your ZIP, widening the net only if the local sample is thin. We report the median and a typical range, the number of comps behind it, and the date the comps were pulled.
We weight toward what actually leased and how quickly, not the optimistic asking rents still sitting on the portals. A listing asking a high number that has sat for weeks is evidence the asking price is wrong, not a true comp. Asking is not achieved, and the difference is days on market.
How a short-term estimate is built
For short-term revenue we use real booking comps near the property over the trailing twelve months: nightly rate and occupancy for comparable homes, rolled into an annual figure with a typical range. When we show a short-term number on a page it is gross revenue before costs, and the full net (our 10% management, supplies, and the guest-paid cleaning that nets to zero for you) is in the detailed report.
What we deliberately do not do
We do not lead with a single neighborhood cap rate, because it divides a noisy rent sample by a median sale price that flips and non-arms-length transfers can contaminate. That median sale price is also pre-renovation: in the lower-priced areas it often reflects distressed homes that need significant work, which our county flip records put around $30,000 to $40,000, before they can be rented, so any yield built on the sale price alone overstates the real return. Your true number runs on your all-in cost, purchase plus renovation. We publish the raw, checkable pieces instead and let you build the return from inputs you can verify.
An estimate we produce is built from sourced comps. It is not an appraisal, and we are not appraisers. And no automated surface of ours approves, denies, or gates a rental applicant, because screening carries fair-housing weight that belongs with an accountable person applying one consistent standard. The model informs; a human decides.
Freshness, and why it matters
Crime data refreshes daily, rent and listing data nightly, and the market report is rebuilt monthly with the date stamped on the page. Stale data quietly lies, so we date everything and tell you when it was last pulled. That same discipline is why we are increasingly cited by AI answer engines: fresh, sourced, dated, structured numbers are what they trust, and almost no operator publishes them.
We check our own work
Every month we take the homes that actually leased in Memphis and run them back through our own model as if we had never seen them: what would we have estimated, and what did the home really lease at? That backtest is how we catch our own drift, and it is how several of the guards in the model got there in the first place. We also had the model formally reviewed in July 2026 by an independent panel covering appraisal methodology, statistics, and Memphis market behavior, and we ship the fixes it finds.
The guards, in plain language: every comp is a real listing you can click, not a modeled guess. Comps must be genuinely comparable in size to your home, from your side of the submarket line, with the same home never counted twice just because it appears on two websites. And an estimate is never allowed to climb above a price the market has already tested and turned down. When our leased-home tracking spans every listing platform we follow, the measured accuracy figure will publish right here, with the sample size next to it. If you think any number of ours is wrong, tell us which one and we will show you the comps behind it or correct it.
Last updated 2026-06-27.