How to Run Comps on a Georgia Investment Property

Four things: proximity, recency, size, and condition. A sale two streets away last month in similar condition tells you something. A sale three miles away two years ago, gutted, does not — even though both are technically "recent nearby sales."

Condition is the one investors most often get wrong, and it is the one that moves the number most. A renovated resale and an unrenovated fixer on the same street are not the same data point. If you are estimating what a house will be worth after you renovate it, an unrenovated sale is evidence about the wrong thing entirely.

Size matters because the arithmetic runs on price per square foot. Take the comparable sales, work out what each one sold for per square foot, and multiply by your property’s square footage. That is the whole method. Everything else is deciding which sales get to be in the set.

Whatever strategy you are running, the sequence is the same. First, build a pool of candidate sales near the subject. Second, filter that pool by something specific to your strategy — this is the step that changes. Third, rank what survives by how closely it resembles your property. Fourth, keep the most comparable few. Fifth, take a central price per square foot from those and multiply by your square footage.

Fundry keeps the top five and starts its search at a one-mile radius, widening to a maximum of three miles automatically when there is not enough nearby evidence. You can pin a wider radius by hand on a sparse rural subject, but the automatic walker stops at three.

It also trims the extremes before averaging — ten percent off each end — so a single outlier cannot drag the estimate. With five values that trim has nothing to remove, which is a reminder that on a thin set every comp carries real weight.

This is the part most tools skip, and it is the difference between a valuation and a number. A flip, a ground-up build, and a rental are three different plans for the same address, and each is measured against a different set of sales.

For a flip, the question is what the house sells for once renovated. So the pool excludes unrenovated sales — fixers and homes needing updating — and it excludes new construction, because a brand-new house is not evidence about a renovated 1960s ranch. It also caps recency at 180 days, because a renovated-resale market moves. Survivors are ranked on square footage similarity, with the more recent sale winning a tie, and the estimate is a trimmed average of their price per square foot. The median is shown beside it as a sanity check: when the two diverge, an outlier is pulling the average.

For a ground-up build, the existing structure is often irrelevant — on a teardown there may not be a meaningful one. So the pool is new-construction sales only, no more than two years old at the time of sale, and they are ranked against the home you intend to build rather than the one standing there now. Ranking blends square footage, bedrooms, and bathrooms. The result is then applied to your planned square footage, not the current footprint. This is the whole point of the strategy, and it is what a generic comp tool cannot do for you.

For a rental, the pool is recent nearby leases together with active rental listings, ranked on square footage similarity, with leased properties preferred over still-active ones. Only the leased rows set the number — an asking rent is what a landlord hopes for, not what a tenant paid. Active listings help rank and fill out the set, but the estimate itself comes from signed leases.

It is natural to assume a closer comp should count for more. It is the most intuitive thing about comps and it is not how Fundry works — distance is not a weight in any strategy.

Distance does three jobs, and none of them is weighting. It sets the radius the search happens inside. It acts as a sort key when narrowing to modern-era sales. And it breaks a tie when two comps score identically on similarity. That is all.

The practical consequence: a comp one block away and a comp two miles away, equally similar in size and equally recent, count exactly the same. If you think the closer one deserves more say, the honest move is not to invent a weight — it is to drop the far one from the set and watch the estimate move.

Thin comp sets are the normal condition in much of Georgia, not an edge case. Rural subjects, unusual homes, and neighborhoods with little recent turnover all produce them.

The right response is to say so. Fundry requires at least two usable comparable sales; below that it reports insufficient comps and shows an empty state rather than a number. A confident-looking estimate built on one sale is worse than no estimate, because you will act on it.

The same logic applies to rentals. If the most comparable five contain no leased property, there is no signed-lease evidence to average, and no rent estimate is produced. An empty result is information: it tells you this address needs a phone call, not a spreadsheet.

An automatically selected comp set is a starting point, never an answer. You know things the data does not: that the sale on the corner was between family members, that the one backing the highway is not really comparable, that the renovation on the third one was cosmetic.

So the last step is judgment. In Fundry, toggling a comparable sale out of the set recalculates the resale estimate, the profit, and the margin immediately, and your curated set overrides the default five. The number moves in front of you, which is the point — you can see exactly how much of your valuation was resting on the comp you just doubted.

Illustrative example. If four renovated sales work out to $195, $205, $215, and $225 per square foot, the average is $210, and a 1,600-square-foot subject estimates to $336,000. Drop the $225 sale as not truly comparable and the average falls to about $205 — roughly $328,000. An $8,000 swing from one judgment call is normal, and it is the reason a valuation you cannot inspect is not worth much.