This is the working agreement for what is worth pursuing. Prefer numbers you can stand behind. If you are using an assumption (a rent you think you can get, a repair budget), write it down as a guess so later you know what was modeled vs what was verified.
The AI assistant is stricter: it will not invent a rent, tax, or repair number. If it does not have a source, it leaves the field blank. You can still type your own assumption and run the math.
It depends on the play:
| Play | How picky we are | We keep looking when |
|---|---|---|
| Buy-and-hold rental, or short-term rental | Moderately picky | After tax / HOA / rent (or nightly rate + occupancy), it still cash-flows. Rough floor if we cannot run a long-term hold: about $100 per month. A fat upside is nice, not required. |
| Fix and flip, or wholesale | Very picky | There has to be a clear spread, not a break-even. Ballpark: flip profit about $40–50k+; wholesale spread about $25–40k+. Thin deals fail. |
If you say “be aggressive” or “be conservative” for a session, we tighten or loosen that bar for everything we look at in that run.
Even if the numbers clear the bar, we do not mark a deal as “pursue” until it also fits at least one investor in your book.
“Almost fits” does not count.
Foreclosures and auction houses stay on “still researching” until tax, title, and repairs are believable — even if a buyer looks interested.
When you reopen an old lead, read the comments first so you do not repeat work.
The score needs rent, nightly rate, or after-repair value. Until those are filled in — from a listing, comps, or your own assumption — the deal stays unscored.
Geography is wide open (US-first). Do not assume we only buy in one city. Explain why this market or this house is interesting.