We Ran 3,658 Houston Listings Through the Purchase Test. Fifteen Cleared It.

By Steve Waller · September 10, 2026

What we pulled

Every active single-family listing in Harris County priced between $80,000 and $300,000, at least 1,000 square feet, built between 1950 and 2010. That is 3,658 listings.

Then one test on each of them, the same test whether the buyer intends to flip the house or keep it and rent it out: purchase price against 75% of after-repair value, less repairs, less closing costs. If the asking price sits above that number, the difference is cash out of the buyer's pocket at the table. We ran it at 70% as well.

The two filters we deliberately did not apply

We did not filter on days on market, and we did not filter on the language in the public remarks. Both are tempting and both are wrong.

Screening for as-is, investor special, handyman and fixer feels like it finds the deals. On this set those two filters together cut 3,658 listings down to 40 before a single number was calculated — and the best property in the county was in the 3,618 that got thrown away. It had been listed nine days and its remarks say nothing about investors. A house does not become a deal because the listing agent used a particular word.

What actually narrows the field is arithmetic. Clearing the test means buying at roughly half of after-repair value, so the only listings that can qualify sit at the bottom of the price-per-square-foot distribution. Sort by that, work up from the cheapest, and the screen finds itself.

How the values were built

Median closed price per square foot from sales inside each subject property's own subdivision over the trailing 180 days — 391 closed sales across 41 subdivisions.

A ZIP-code price per square foot is not a comparable set. In northwest Harris and in Fort Bend the ZIP average is carried by newer master-planned sections, while a 1990s subdivision inside that same ZIP trades at close to half of it. We have watched that shortcut produce a 43% valuation error on a single file.

What came back

Fifteen listings clear the test. Two of the fifteen work at the price the seller is already asking, with nothing out of pocket. One of those two clears even at a 70% advance, which nothing else in the county managed. Seven more sit within $30,000 of the asking price, which is a negotiation rather than a wall. The remaining six need a real cut — 16% to 28% off — and several of those sellers have been waiting long enough to listen. One has been on the market 613 days.

Why the other 3,643 fail

This is the part worth reading, because nearly every one of these looked like a bargain in a search result.

• The house is oversized for its street. Six failed on this alone. A low price per square foot usually means the house is larger than anything the subdivision has ever closed, and the after-repair value you would need sits 30% to 75% above the neighborhood's top sale of the last six months. The appraisal does not go there.
• Cheap per foot, but the whole street is cheap per foot. In four subdivisions, large homes themselves trade at $80 to $97 per square foot. A listing asking $80 to $87 there is at market, not below it. There is no gap for the formula to open in.
• The square footage in the MLS is wrong. One listing advertises 6,534 square feet at $37.50 a foot. That figure is the standard lot size for that subdivision and the neighbors are around 2,000 square feet — the lot number is sitting in the building field. After-repair value is a straight multiple of square footage, so that one bad field would have produced a valuation three times too high.
• You would be buying somebody else's finished flip. One 3,716 square foot listing asking $245,800 is the same house — same lot, same footage — that sold as an as-is estate in May of this year for $160,000. The spread has already been taken.
• It is a lot, not a house. One is offered for lot value only, with the adjacent parcel also available. There is no after-repair value to test. That is a ground-up build on two lots, which is a different loan and a better conversation.

What no spreadsheet surfaced

After the arithmetic was finished, reading the listing remarks reordered the results.

Five of the fifteen survivors are foreclosures, and three of those are occupied with no interior access and no inspection contingency. One states plainly that the seller cannot guarantee occupancy. A repair estimate on a house nobody is permitted to enter is a guess, and if the occupant does not leave, removing them is the buyer's cost and the buyer's timeline. One of those three had been sitting at the top of the list on the numbers alone.

The cheapest property per square foot in the entire county names roof, HVAC, foundation, siding, flooring and termite damage in its own remarks. Carrying repairs at a realistic number instead of an inferred one moved it from a $6,750 gap to a $31,750 gap in a single line of the model.

Not every foreclosure is the same animal. A bank-owned property where the lender holds title and the house is empty is workable. An occupied auction is a different transaction wearing similar clothes.

The point of all this

Fifteen out of 3,658 is four tenths of one percent. That is not a reason to stop looking — it is the reason to run the number before writing the offer rather than after the loan closes. By the time financing is in place the purchase price is fixed, and no lender can repair a deal that was bought wrong.

The deal calculator models the purchase, the rehab draw and both exits on any property, and the DSCR calculator handles the rental side if the plan is to keep it.

Want the fifteen addresses?

The screened list — the addresses, the supported purchase price on each, the comparable sales behind every valuation, and the condition and access flags — goes out one to one, not published. MLS data is licensed, and a list like this is worth more to somebody who is actually going to write on one of them than it is sitting on a public page.

Reply to this, or call 936-522-8951, and it comes back the same day. If you already have a property under contract or under consideration, send the address instead and it gets run the same way — the comparable sales, the supported purchase price, and an honest answer about whether it works.

Sources

• Houston Association of Realtors MLS — active listings and closed sales, accessed 10 September 2026

Market data is from public sources and is provided for information only. Nothing here is an offer of credit, a rate quote, or investment advice. Financing is subject to underwriting, lender approval, appraisal or valuation, title, insurance, property eligibility and program availability.

All briefs