Start free

Full transparency

Where your verdict comes from

Every grade on ATD is two layers that never mix: live research — comps, taxes, insurance, today's rates, local rules, each labeled with its source and date — feeding locked math that decides the tier the same way every time. The AI researches; it never grades its own homework. Here is what every strategy actually looks at.

Short-term rental (STR)— graded on: Realistic monthly net after every cost, at researched nightly rates and occupancy.
  • Nightly-rate and occupancy comparables, including listings in the same building
  • The hotel-price ceiling within a mile (your rate can't live above it for long)
  • Regulation with ENFORCEMENT reality — permits, minimum nights, occupancy caps, quiet-hours rules, and the lodging tax (who collects it)
  • 13 risk checks: pests, safety trend, HOA bans, supply density, demand anchors, noise, construction pipeline, and more
  • Seasonality, amenity gaps vs the top competitors (with payback time), host-insurance reality
Long-term rental— graded on: Monthly cash flow after the FULL cost stack, and the return on your actual cash in.
  • Effective market rent — advertised concessions are amortized in, not ignored
  • Post-sale property tax (many places reassess at YOUR purchase price) and today's insurance market, not the seller's old bills
  • Today's mortgage rate for your location, researched at run time
  • 10 hold checks: comp quality, habitability, safety, title/owner record, supply, demand anchors, noise, pipeline, violations, trajectory
  • The rules where the property lives: eviction timeline, rent control, deposit caps, licensing, special-district fees
Section 8— graded on: Rental cash flow on the rent you can actually COLLECT — the lower of the published FMR and the street rent (the housing office's rent-reasonableness test).
  • HUD Fair Market Rent (table-cited) vs open-market comps, payment standards, utility allowances
  • Inspection reality: the unit must pass every year for payments to continue
  • Turnover math — vouchers follow the tenant, and the first check takes 60–90 days
BRRRR— graded on: Capital recycled at the refinance plus BOTH exits — refinance-and-hold or sell — the better door carries the grade.
  • After-repair value from renovated SOLD homes, price trend and days-on-market for the exit
  • The refinance sized the way lenders actually size it: the LTV cap AND the rent-coverage test (whichever gives less), minus the refi's own closing costs
  • Seasoning windows, the delayed-financing exception for cash buyers, appraisal preparation
Fix & flip— graded on: Net profit in dollars AND margin after both closings, holding, and ~8% selling costs — before your income taxes (add your rate to see after-tax).
  • Renovated comps, days-on-market and 12-month price trend (your market timing)
  • Listed square footage vs the county assessor (unpermitted space won't appraise)
  • FHA anti-flip windows at starter prices, vacant-home insurance, lien-waiver discipline, soft-market concession drag
Wholesale— graded on: Your assignment fee, capped by the real spread — the contract must leave the end buyer room to profit.
  • The most-you-can-offer math worked backwards from the end buyer's numbers
  • Cash-buyer activity and resale speed in the zip
  • Your state's assignment disclosure/licensing rule, earnest-money norms
Subject-to— graded on: Cash flow on the seller's inherited loan AND survival of the called-loan scenario.
  • The inherited payment vs a today's-rate loan (the due-on-sale swing, priced)
  • Equity build on the old amortization schedule, servicing/escrow structure flags
Buy to live in— graded on: Own-vs-rent premium plus the 5–10 year picture — projected equity from local appreciation history and loan paydown. A home is a decade decision, not a year-one return.
  • Price fairness vs sold comps, local 5–10 year appreciation history
  • Verifiable decade drivers: school direction, FUNDED transit, employer moves
  • Value-liability features for your climate — pools in cold regions, leased solar, buried oil tanks, over-customization
  • Government first-buyer programs and the tax angles of your jurisdiction (US and Canada)

The honesty rules the math lives under

  • Locked, deterministic grading — research informs it, never overrides it.
  • Every researched figure carries its source and as-of date on the verdict.
  • No data means NO DATA — estimates and fallbacks are labeled, never disguised.
  • Check counts are computed from your actual run — never inflated.
  • Profit is shown pre-tax and says so; add your rate to see after-tax.
  • Decision support, not financial/legal/tax advice — always do your own due diligence.

The accuracy scorecard

Every analysis is grounded on independent industry anchors (licensed and official feeds), and we grade ourselves against them continuously — from real runs, not a demo set. Numbers below cover the last 30 days and refresh hourly.

5

analyses grounded on live anchors

5.2

independent anchors per analysis (avg)

—

rent checks within 10% of the independent AVM/FMR

0

disagreements flagged to the customer (the check working)

Early days for this scorecard — 5 measured runs since the anchor layer shipped. The sample grows with every analysis; we publish it anyway because that is the point. Anchor sources in this window: AirROI Markets API · Bank of Canada · V80691335 · FEMA NFHL · FHFA HPI via FRED · FLSTHPI · FHFA HPI via FRED · OHSTHPI · FHFA HPI via FRED · PASTHPI · FRED · MORTGAGE30US · HUD FMR API · HUD Income Limits API · RentCast AVM · Walk Score API. A flagged disagreement renders as a CAUTION on the verdict — we surface it, we never average it away.

The Questions library

Real questions, straight answers. New ones get added as you ask them.

Where does my verdict actually come from?

Two layers that never mix: live research (comps, taxes, insurance, rates, rules — every figure labeled with its source and date) feeds locked, deterministic math that decides the tier and verdict. The AI researches; it never grades its own homework. The same inputs always produce the same grade.

Can the AI make up numbers?

Research must cite a source for factual claims or mark the item NO_DATA — a missing number is shown as missing, never invented. When we use a fallback (like a default rate), it's labeled as a fallback right on the verdict, with a prompt to enter the real figure.

Why does the check count change (13 vs 9–10)?

Each strategy runs the checks that apply to it. STR runs all 13; purchase strategies mark the STR-only checks not-applicable and the verdict counts only what actually ran — the number is computed from your result, so it can never overstate.

Why did the same address grade differently under two strategies?

Because the strategies measure different things in different units: a flip is graded on project profit in months, a rental on monthly cash flow, a home on own-vs-rent over a decade. One address can be a strong flip and a weak rental at the same price — that's the point of strategy-native verdicts.

Is this financial, legal, or tax advice?

No. ATD is decision support — researched facts plus transparent math. Every report carries it: always do your own due diligence, and confirm rates, taxes, insurance, and legal questions with your lender, accountant, and attorney. We show our work precisely so your professionals can check it.

Do you have to disclose your methodology?

There's no law requiring a decision-support tool to publish its method — we do it because trust is the product. What we DO owe you, and deliver on every run: the sources behind each researched figure, honest labels on every estimate and fallback, and grading rules that don't move between runs. The full recipe-level detail (our exact prompts and thresholds tuning) stays internal — that's the craft — but nothing about where your numbers came from is ever hidden.

What do deep runs add, and what do runs cost?

A standard run researches and grades the deal. A Deep run (5 credits) adds live comp validation — real listings, review themes, pricing rate cards, regulatory dossier — the difference between a model and a model checked against the street. Credits are listed on the pricing page; the first 10 deals are free, no card.

What data sources does ATD use?

Live web research across listing platforms, county records, HUD tables, city portals, and news — plus licensed data feeds we're expanding (rental AVMs and STR market data are next). Every integrated source appears as a labeled input on your verdict with its as-of date. We never silently blend sources.

Why does my verdict warn me about things I didn't ask about?

Because the things nobody asks about are what burn people: post-sale tax reassessment, vacant-home insurance exclusions, rent-reasonableness caps, refi coverage tests, pool upkeep in cold climates. The operator layer prices in what 5–10 years in each trade teaches — so your first deal doesn't have to.

I found something wrong or have a question you haven't answered.

Send it through the Feedback button on any page or the contact page — real questions get answered and the good ones get added to this library. If research got a fact wrong, tell us: correcting it improves the next thousand verdicts.