You can run accurate comps without MLS access, without a license, and without leaning on an agent every time. You need the right sources, a disciplined way to adjust for differences between properties, and an honest read on where free data runs out. That last part matters most, because when the data is wrong and you trust it anyway, you overpay for the house.
A comp is one recent sale that matches your property
A comp is a single recently sold property similar to yours in location, size, condition, and features. A comp set is the group of them you pull together, three to six sales from the last three to six months, as close and as similar as possible. The MLS shows actual closed prices, which is why it's the cleanest source. You won't have access to the MLS without a real estate license, but in most states that same closed-price data flows directly into public sites through MLS data feeds.
Step 1: Pick comps that actually hold up
Not every nearby sale is a real comp. Stay within about half a mile, in the same neighborhood or a comparable one, and don't cross a boundary that changes value, a school district line or a busier road. Match property type, stay within roughly 20 percent of your square footage, and keep bed and bath counts close. Favor the most recent sales, since older ones drift from current conditions. The tighter your set matches the subject, the less adjusting you'll do later, and the more defensible your number is.
Step 2: Pull sold data from three sources
| Source | What you get | Limitation |
|---|---|---|
| Zillow / Redfin | Sold price, photos, listing history | Data lags county records by days |
| Realtor.com | Sold price, MLS-sourced listing detail | Coverage gaps in smaller markets |
| County assessor / tax records | Recorded sale price, no account needed | Clunky interface, lags closing by days to weeks |
On Zillow or Redfin, filter for Sold, set the date range to the last 90 days, and sort by distance from the subject. The photos let you match condition and finish, not just bed and bath count, and when the photos line up well, the number lands close to a formal appraisal. Then confirm against the county assessor or tax record site, which shows the recorded sale price for most transactions with no account needed. Some sales show up only in the public record and never hit the listing sites, and those tend to matter more when you're pricing as-is value rather than after-repair value. Where the portal and the county record agree, trust the number.
Step 3: Adjust each comp for what it's missing
Price per square foot is the simplest disciplined method. Average the sale-price-per-square-foot of your comps, then multiply by your property's square footage. Comps at $200 a foot and a 1,500-square-foot property put you at $300,000 as a starting value.
Then adjust for what square footage misses: bed and bath count, garage, basement, lot size, and condition. Each difference carries a dollar value. Skip that step and you land too high when your comps are nicer than your property, too low when they're worse.
Where the free method stops working
The whole approach depends on sale prices being public. In twelve states, they aren't. Assessor records there show an assessed value that rarely matches market price, and portal estimates sit on incomplete data. Investors in those markets learn this the hard way, usually after relying on a number that was never backed by real sales. These include some of the largest wholesale markets in the country, not an edge case. Free public records can't solve that. You need a source with the actual sales behind it.
For which states hide sale prices and what to do instead, see Non-disclosure states for real estate investors.
Where the manual method hits its ceiling
Matching photos and weighing a property against six comps works fine for one deal. Running fifteen a week while a seller waits on you is where it breaks. ChatARV pulls from the same sources, Zillow, Redfin, and county records, matches and adjusts them automatically, and returns an after-repair value and an as-is value in about 60 seconds, including in the non-disclosure states where none of this data exists publicly to begin with.
FAQ
How many comps do I actually need for a defensible number?
Three is the minimum, six is comfortable. Fewer than three and one bad comp skews the whole set. More than six rarely adds accuracy, it just adds time spent adjusting sales that were already weaker matches to begin with.
What if I can't find three good comps within half a mile?
Widen the radius before you widen the date range. A comp from a mile away that closed 30 days ago beats one from your block that closed eight months ago, unless the extra half mile crosses a real value boundary like a school district.
Do sold comps expire?
Not on a fixed timeline, but weight them like they do. A 90-day-old comp in a stable market still holds up. In a fast-moving one, anything past 45 to 60 days needs a downward adjustment for market drift before you use it.
Run comps without MLS access
ChatARV matches and adjusts sold comps automatically and returns an after-repair value and an as-is value in about 60 seconds, including in non-disclosure states.