How every number is made

Most tools take the asking price and grade it. Faro runs it the other way. You state the return you require, a cash-on-cash floor, a DSCR minimum, a minimum monthly cash flow, whatever your buy box holds, and the solver returns the highest purchase price at which every one of those rules still passes. Knowing a deal is bad doesn't help; knowing the number where it stops being bad does.

It is deterministic arithmetic, not a model estimate. The solver walks the price axis against the same underwriting engine that produces the rest of the report, so the target price and the cash flow shown beside it can never disagree. Reproduce it by hand from the inputs on screen and you'll get the same number.

The target-price curve

The report also charts the deal across every price, shaded where your rules pass and where they break. Your walk-away number is where the colour changes. That curve is allowed to be red all the way across. When no price clears your rules, that is the answer, and manufacturing a green light instead would make the whole tool useless.

Everything is an assumption you can move

Rate, rent, vacancy, capex, insurance, down payment, holding period, and each buy-box rule are editable, and the target price re-solves live as you move them. Property tax and insurance defaults are state-aware rather than a national rule of thumb that's wrong nearly everywhere. Defaults are a starting point, never a verdict.

The deal score is a 0–100 read on the property at its asking price not at your target price, which is a different question the solver already answers. It combines the returns the engine computed, how the property sits against local rents and values, and a risk adjustment, then reports the result with a confidence band rather than as a single confident-looking integer.

City deal tiers

"A good deal" is meaningless without a market to measure it in. Faro carries deal-tier benchmarks, weak, fair, average, good, and great, for 19,464 cities, calibrated against local price and rent distributions. A good deal in Cleveland and a good deal in San Francisco are each measured against their own market rather than a number borrowed from a podcast. Where a city benchmark would loosen the state-level default it is clamped: a benchmark may tighten your target, never relax it.

210 of those cities also have a full market page with the current tier prices and rents. Browse the markets →

Risk adjustment

Scores are risk-adjusted on natural-hazard exposure, so a cheap property with a high hazard reading can't top a ranking purely on price. The hazard inputs themselves are described in the hazard section below.

Every valuation carries a confidence badge, and it is tied to how much comp depth the analysis actually had not to how good the deal looks. Thin coverage widens the band. It never quietly narrows it to make the report read as more certain than the data supports.

How many comps a run actually gets

Each analysis pulls up to 15 sale comps and 15 rental comps, shown as cards and on a map so you can see exactly which properties drove the estimate, and throw out the ones that don't belong. Across stored production reports a typical run comes back with the full 15 of each; runs with materially fewer happen, but they are rare, and they are exactly the ones the confidence band is there to flag.

The comps are similarity-selected, not merely nearest, and the estimate is a weighted average rather than a simple mean, so a single outlier sale can't drag the valuation.

Two kinds of comp, weighted differently

An analysis reads recorded sales what buyers actually paid, from public records, and active listings, which are what sellers are currently asking. They answer different questions and Faro does not treat them as the same evidence. A paid price is a fact; an asking price is an opinion. But closings lag, and in a moving market a year-old sale describes a market that no longer exists, while asks are current and run systematically above what homes fetch.

So closings anchor the level and asks adjust it: the buyer-side estimate weights recorded sales at 65% and active asks at 35%. Within each set, every comp is weighted by similarity, distance and how recent it is. When a property has only one kind of comp, that set is used on its own terms rather than blended with nothing, and the report says which evidence stood behind the number.

The seller-side view inverts it, deliberately. What you can ask is a question about the homes a buyer is choosing between right now, so there the current listings carry the larger share.

What Faro won't do

It won't invent a comp when the data is thin. It lowers the confidence badge and says so. It won't hide the assumptions behind a result; the drawer opens on every report. It won't tell you a property works when the arithmetic says otherwise, which is the entire reason the target-price curve is allowed to be red all the way across. And it won't present an AI-refined figure as if it were computed. Those are labeled where they appear.

Why not a spreadsheet

Spreadsheets are fine for one deal a quarter. They break down when you're comparing six properties across three markets and the rent assumptions came from memory.

Every Faro analysis carries a wildfire and a flood risk score on a familiar 1–10 scale, derived live from public-domain U.S. federal data the FEMA National Flood Hazard Layer and the USDA Forest Service's Wildfire Risk to Communities program. As works of the U.S. government these datasets carry no copyright and are free to use. We query the official layers directly for each address, so the reading reflects the current effective maps: not a stale copy.

The scores are estimates for screening, meant to make two properties quickly comparable. They are not official FEMA flood-insurance determinations or certified wildfire assessments.

The wildfire score

1. Locate the home precisely from its address.
2. Sample the surrounding area, not just the lot. We read the fire environment across a neighborhood around the property, roughly a one-kilometer radius, not a single point. This is what captures wildland-urban-interface exposure: developed lots often model as low-risk on their own footprint, yet burn from embers carried out of the wildland around them, which is how towns like Santa Rosa and Paradise were lost.
3. Translate to 1–10. The sampled risk is mapped onto Faro's calibrated scale, so the number reads the way a buyer expects.
4. Apply a fuel-hazard floor. A second reading from the Forest Service's hazard-potential layer sets a minimum, so a home surrounded by fire-prone terrain can't score artificially low even when its own parcel is built up.
5. Band the result from Very High down to Minimal, or "Not rated" where no data covers the point.

We validated the scores against real communities with known wildfire history: high-loss towns land in the High-to-Very-High range while dense urban cores read Minimal: confirming the number tracks real-world outcomes, not a single map pixel.

The flood score

The flood score reads the FEMA flood zone the property's exact location falls in, the same zones that drive federal flood-insurance requirements, and places it on the 1–10 scale by regulatory severity:

Coverage

Both scores resolve nationwide, every state, including Hawaii and Alaska, for any geocoded address in a populated place. The wildfire data spans the full national extent, and FEMA's flood maps cover essentially all populated areas. Where FEMA has genuinely not mapped a location, the flood score reads "Not rated" rather than guessing.

Limitations & honesty

• Every score is an estimate off the official layer, a screening signal, not a certified determination, and won't be identical to any other provider's number.
• The wildfire reading is a worst-of-neighborhood view, so a home on an isolated safe lot beside steep, fire-prone terrain reads closer to that terrain. That's deliberate, it's how those homes actually burn, but it's a conservative bias, not a per-structure defensible-space audit.
• Flood scoring reflects the mapped FEMA zone only; it does not model rainfall (pluvial) flooding outside designated zones or future climate change.
• Both are only as accurate as the address geocode.

In short: verify with FEMA, the USFS, and your insurer before relying on any of it. Faro's job is to surface the risk early, not to replace an official determination.

A market scan can screen vacant land as well as rental property, and when it does it switches models entirely. A parcel has no rent, so it has no cash flow, no cash-on-cash return, no debt coverage and no cap rate. Those figures are absent, not zero and a land listing run through a rental model produces a confident, invented rent rather than an obvious error. So land is filtered out of the rental scan and scored on its own terms.

Land is priced by parcel, not by the acre. Across a full scan of one city's active land listings, the median price per acre fell roughly eight-fold from small infill lots to twenty-acre blocks. Comparing a parcel to a single market-wide price per acre would rank purely by size and call it a deal signal, so every parcel is compared to what its market pays for a parcel that size.

The score is how far below that local expectation a parcel is priced, measured in that market's own spread rather than in a fixed percentage. Land prices are dispersed enough that a 35% discount can be ordinary noise in one market and a real find in another. Buying at the local expectation scores in the middle of the range, not at the bottom.

What we check on a parcel

• Buildability, from free public federal sources: mapped wetlands (U.S. Fish and Wildlife Service), hydrologic soil group and drainage class (USDA NRCS), what a county health department reads when deciding whether a septic system is permittable, ground slope and elevation (USGS), and designated critical habitat for listed species (USFWS).
• Hazards: FEMA flood zone and USFS wildfire hazard, the same layers used elsewhere on this page. These weigh more heavily on land than on a house. A dwelling in a flood zone is an insurance premium, while a parcel in one may simply not be buildable.
• Carrying cost: property tax against an asset that produces no income while you hold it.
• Time on market, calibrated to land. Land sits; a parcel listed for four months is ordinary, where a house at four months is a signal.
• Recorded sales of nearby land, when a parcel is opened in full: actual closing prices rather than asking prices, each adjusted to the subject's size.

Those findings lower the score a parcel you may never be allowed to build on should not outrank one you can, and they are also listed individually rather than rolled into a single figure. A wetland is a permitting problem, poorly-drained soil is a septic problem and a steep slope is a foundation problem: different costs with different remedies. The score says how much the constraints matter; the detail says which one you are facing.

They are weighted by how hard the constraint is to remove, not by how alarming it sounds. A steep slope is an engineering cost: real, knowable, solvable with money. Wetland permitting or an endangered-species consultation is a regulatory outcome nobody can promise, and an uncertain answer is treated as worse than an expensive one.

What land scoring cannot tell you

These gaps matter more on land than anywhere else on this page, because any one of them can decide whether a parcel is worth anything:

• Utility service. Whether water, sewer or power can actually be brought to a parcel is a question for the utility, not a public dataset. A serviced lot and a raw one score the same here.
• Road frontage and legal access. A parcel can be landlocked without any public record saying so.
• Easements and title. Not public data. That is what a title search is for.
• Zoning comes from the assessor record where one exists, and assessor records lag. It is a starting point for a call to the county planning department, not a substitute for one.
• Comparable land sales are few and old. Land turns over slowly, so nearby sales are often several years old. We show each sale at its recorded price and date and state how old it is, rather than adjusting it forward by an assumed appreciation rate we cannot support.

Land screening is for narrowing a list. Before committing to a parcel, the county planning department, a title company and a surveyor answer questions no public dataset can.