Every real estate underwriting platform makes decisions about which comps to show you. Most of them don't tell you exactly how they make those decisions — or they use a single blended score that's hard to interpret or challenge.
Underwritey ranks comps in tiers, and the tiers are explicit. Here's exactly how it works, and why we designed it this way.
The ranking logic
When you enter an address, Underwritey pulls recent sold transactions and ranks them in a defined sequence:
Tier 1 — Location and recency. A sale two blocks away that closed last month ranks higher than a sale half a mile away from eight months ago, even if the second house is a better property match. We weight these two factors first because they capture the most important truth in comp selection: you're pricing into a specific local market at a specific moment in time.
Location is evaluated by distance from the subject. We use a radius approach, with closer properties ranked higher within the same recency band.
Recency is evaluated by days since closing, with most recent ranked first.
Tier 2 — Property fit. Within the closest, most recent sales, we then rank by how well the property matches your subject: beds, baths, square footage, and year built. A 3/2 that's 1,400 sqft and built in 1985 ranks higher against a similar subject than a 4/3 that's 2,100 sqft and built in 2005.
Tier 3 — Data quality signals. We apply several filters to identify likely distressed, non-arm's-length, or otherwise atypical sales that shouldn't drive ARV. Properties that appear to be related-party transfers, sales with DOM of zero or one, or sales that are statistical outliers in the local context are ranked lower or flagged.
Why location and recency come before property fit
This is a deliberate choice, and it differs from how some automated valuation models work.
Some AVM systems prioritize property similarity above everything else — finding the "most similar" house regardless of how far it is or how old the sale is. The problem is that real estate markets are hyperlocal and time-sensitive. A perfect property match two miles away in a different school district is a much weaker data point than a good-enough property match three blocks away that sold last month.
The investors who get into trouble on ARV are usually the ones who picked better-matching comps at the expense of location and recency. The appraiser on their BRRRR refinance, or the buyer evaluating their wholesale deal, applied location and recency filters first — and the ARV gap appeared.
You're still the underwriter
This is the part that matters most. The automated ranking is a starting point, not a conclusion.
You see every comp we've ranked. For each one, you can see the address, the sale price, the sold date, the beds/baths/sqft, and the distance from your subject. You can also see why it ranked where it did.
You can remove any comp with one click. If a ranked sale was a foreclosure, a family transfer, an outlier, or simply a house you know doesn't reflect the real market for your subject, remove it. The ARV recalculates immediately using only the comps you kept.
You can also add comps manually if you found a sale we missed or if you have MLS access to additional data. Any comp you add gets the same treatment as the ones we pulled — it shows up in the ranked list and contributes to the ARV.
Why removing bad comps matters more than having more comps
There's a temptation to include more comps to feel more confident. More data points should mean more accuracy, right?
Not always. One bad comp can materially distort your ARV. A short sale at $40,000 below market value, included in a pool of 6 comps, can pull your median down by $6,000-$8,000. That's a meaningful error on a deal where your MAO depends on an accurate ARV.
By keeping the removal process easy and instant, we make it easy to stay conservative. If you're unsure about a comp, remove it and see what the ARV looks like without it. If the number barely moves, the comp wasn't doing much work. If it moves significantly, you've just learned something important about how influential that comp was.
What the ARV represents after your review
Once you've reviewed the comps and removed any that don't belong, the displayed ARV is a median-based calculation of the comps you've chosen to include. It's your ARV, supported by comps you've verified. The math is visible so you can explain every number to a partner, lender, or buyer.
That traceability is what makes the number defensible. "ARV is $215,000 based on four comparable sales within 0.3 miles, all sold in the last 65 days, all renovated 3/2s" is a number you can take to a lender, share with a cash buyer, or use in a presentation to a motivated seller explaining your offer.
Start your first underwrite on any property — we'll pull the comps and you can see the ranking for yourself. Try it free →