There's a version of comping a house that sounds complicated — adjustments, paired sales analysis, cost approach — and a version that working investors actually use. This guide covers the version that works in the field, on a time budget, and holds up when a buyer or lender asks how you got your number.
The core question you're answering: what would this property sell for in its fully renovated condition, listed on the open market today? Everything else is how you answer that question with evidence.
The three tiers, in order
Experience teaches investors that comp quality comes down to three things, and the order matters:
Location is always first. Two houses that are identical on paper can be worth completely different amounts if they're in different school districts, on different sides of a highway, or in neighborhoods with different price trajectories. Start by drawing a circle around your subject property — half a mile for most urban and suburban markets, up to a mile for rural areas — and stay inside it.
The tricky part of location is learning where the invisible lines are. A four-lane arterial road, a railroad track, a school boundary, a flood plain — these all create market separations that don't appear on a simple map but absolutely show up in prices. When you're not local, look at where sale prices drop or jump sharply. That's the boundary.
Recency is second. The market you're pricing into is the market today, not the market from two years ago. For most purposes, use sales from the last 90 days. If the market is thin and you can't find enough in 90 days, extend to 6 months — but note that the older sales are weaker evidence, especially in a moving market.
Property fit is third. Within the sales that pass location and recency filters, narrow by similarity: beds, baths, square footage (within about 15-20%), year built (within roughly 20 years), and condition. The condition piece matters most for ARV work: you want comps that sold in a similar finished state to what your property will look like after renovation, not as-is sales of distressed properties.
What actually makes a bad comp
Investors who get burned on ARVs usually made one of a few mistakes:
Using listing prices instead of sales. This is the classic error. What a seller is asking for tells you nothing useful. What a buyer actually paid tells you everything. Sales only.
Reaching across a market boundary. The comp two blocks away across a highway looks fine on a map. It's in a different school district with a different price floor. That half-mile radius exists for a reason — don't compromise it for convenience.
Mixing renovated and as-is sales. If you're estimating ARV — what the property will sell for after renovation — you need comps that sold renovated. Using as-is sales systematically underestimates ARV. Using a mix gives you a blended number that doesn't represent either thing accurately.
Outlier sales that aren't arm's-length. Short sales to related LLCs, estate sales that went way under market, foreclosure redemptions — these sales don't represent what a willing buyer and seller negotiate on the open market. Exclude them. Red flags: sold price significantly below surrounding comparables, DOM of zero or one, buyer and seller share an address, cash sale with no arm's-length disclosure.
Size gaps that are too large. A 1,100 square foot house doesn't comp cleanly to a 1,700 square foot house without significant adjustment. The buyers are different, the renovation costs are different, the buyer pool is different. Stay within 20% on square footage or justify the gap explicitly.
Price per square foot: how to use it right
Price per square foot is useful as a sanity check on your adjusted comp value, not as a primary method. Here's how to use it:
Take the sold price of each comp and divide by its finished square footage. Calculate the median PSF across your best 3-5 comps. Multiply by your subject's finished square footage.
If that number is close to your adjusted value (within 5-8%), your comps are probably well-matched. If there's a bigger gap, something is off: either one comp is an outlier pulling the PSF away from the others, or your subject has an unusual characteristic (very small lot, finished basement counted differently, garage with ADU) that doesn't PSF-normalize well.
PSF works best in neighborhoods with relatively homogeneous housing stock. It works poorly when houses vary a lot in layout, lot size, or finish level.
What to do when you can't find enough comps
In thin markets, rural areas, or unusual property types, you sometimes can't find three solid comps within half a mile from the last 90 days. When that happens, in order:
Extend the time window first. Go from 90 days to 6 months, then to 12. Note the trend in prices during that period.
Expand the radius second. From half a mile to three-quarters, then a mile. As you expand, note how different the market characteristics are at the edges.
Match properties more loosely third. If you can't find 3/2 houses, use 3/1 and 3/3 comps with explicit adjustments for the bath count difference.
Use lot PSF comps fourth. In rural markets or areas with large lot variations, price-per-lot-acre can supplement price-per-sqft.
Talk to a local agent. When data is thin, local expertise fills the gap. A good agent who closes a lot of deals in that specific neighborhood will know things the data doesn't show.
Always document your process when comps are scarce. "I used sales from 8 months ago because there were no qualifying sales in the last 90 days; the trend over that period showed 2% appreciation" is defensible. An unsourced ARV is not.
How adjustments work
Your best comps will still differ from your subject in some ways. Adjustments move the comp's sale price toward what it would have been if it were identical to your subject.
If a comp has a garage and your subject doesn't, you'd subtract the value of that garage from the comp's sale price. In most markets, a single-car attached garage adds $8,000-$15,000 in value. So a comp that sold at $240,000 with a garage might adjust to $227,000 when compared to your garageless subject.
Common adjustments investors use:
- Extra bedroom: $8,000-$20,000 depending on market
- Extra bathroom: $5,000-$12,000
- Garage per stall: $8,000-$15,000
- Finished basement (per sqft): $30-$50
- Pool: $10,000-$25,000 in warm climates; sometimes negative in cold climates
- Size difference (per sqft over/under): median PSF × sqft difference
Keep adjustments small relative to sale price. If you're adjusting $50,000 on a $180,000 comp, you need a better comp — not bigger adjustments.
Putting it together
Your ARV is your professional estimate of where this specific property lands relative to the comparable evidence. It's not a simple average of your comp prices. It's a weighted judgment that accounts for how close each comp is to your subject and in what direction the adjustments go.
For underwriting purposes, be conservative. If your comps cluster around $225,000-$235,000 and your subject has a smaller lot than most of them, $220,000 might be the right ARV. That conservatism is what keeps your MAO from getting you into trouble when the market softens or the renovation runs long.
Underwritey pulls comparable sales automatically, ranks them by location and recency first, and shows you the ARV math. You can remove any comp that doesn't belong and the value updates instantly. No data entry, no formula building — just you deciding which sales tell the real story. Start your first underwrite →