Help

Glossary, verdict rules, underwriting assumptions, and data refresh — all rendered from the same source the code runs.

Glossary

The same data behind the dotted-underline terms across the site

ARV

After-Repair Value

The price this home should sell for once renovations are complete. Estimated from nearby recently renovated sales (comps).

Example: Adjusted comp average $1.03M → valued at $960K after the risk haircut.

Comps

Comparable Sales

Similar homes that sold recently, close by. They are the entire basis of the valuation — the fewer the comps, the less trustworthy the estimate.

Example: 12 similar sales within 0.8 mi in the last 6 months → strong evidence.

DOM

Days on Market

How long the listing has been active. The longer it sits, the harder it is to sell — and the more likely the seller accepts a lower offer.

Example: 92 days on market (city median 34) → negotiation leverage.

$/sqft

Price per Square Foot

The standard unit price for US homes. Comparing $/sqft against nearby sales shows whether a listing is priced high or low for its area.

Example: $612/sqft vs. a ZIP median of $640/sqft → about 4% below market.

Target Buy

Target Purchase Price

The model's suggested offer: the price at which the deal still earns a reasonable profit after rehab and all costs. Usually below list price.

Example: Listed at $798K, target buy $745K (6.6% below list).

Gross Spread

Gross Spread

Sale price minus purchase price, before subtracting rehab, holding costs, or commissions. Gross spread is NOT profit.

Example: $260K gross spread → roughly $110K net profit after all costs.

Net Profit

Net Profit

What is left after subtracting the purchase price, rehab, holding costs, loan interest, and both sides' transaction costs (pre-tax).

Example: Base-case net profit $68K (conservative -$12K to optimistic $114K).

Rehab Estimate

Renovation Budget Estimate

Estimated from square footage × historical renovation costs for similar flips — it is not a contractor quote. Treat it as an order of magnitude until you have on-site bids.

Example: 1,450 sqft medium rehab ≈ $95K (±$30K).

Deal Score

Composite Deal Score

A 0-100 ranking score blended from rules and machine-learning models. It is for sorting, not a probability of profit; 80+ is roughly the top 10%.

Example: Score 86/100 → higher than about 90% of active listings in the city.

Evidence Grade

Evidence Grade

How hard the valuation evidence is, combining comp count, distance, recency, and confidence: Strong / Moderate / Weak / Insufficient.

Example: 12 comps · median 0.8 mi · confidence 78 → Strong.

ARV Confidence

Valuation Confidence

0-100, how confident the model is in its after-repair value. Below 40 the model abstains entirely (gray light).

Example: Confidence 78/100 → usable; confidence 35 → Insufficient Data.

Haircut

Valuation Risk Haircut

A deliberate discount applied to the after-repair value — the fewer the comps and the older the home, the bigger the cut. Better to understate the sale price than overstate your profit.

Example: Raw comp value $1.03M → 12% haircut → underwritten at $960K.

70% Rule

The 70% Rule

A flipper's rule of thumb: purchase price + rehab should not exceed 70% of the after-repair value, leaving 30% to cover costs and profit.

Example: ARV $1M → keep purchase + rehab under $700K.

Negotiation Leverage

Negotiation Leverage

How likely the seller is to accept a lower offer, based on days on market, price-cut history, and how often nearby sales close below list.

Example: 92 days on market + a 4% price cut already → high leverage.

Retail-Ready Risk

Already-Renovated Risk

Photos or the description suggest the home has already been renovated — meaning the flip upside may already be baked into the list price.

Example: Description mentions "turnkey, new kitchen" → verify with photos.

Holding Cost

Holding Cost

Property tax, insurance, utilities, and loan interest while you own the home. The longer the project runs, the higher it climbs — the cost novices most often underestimate.

Example: 6-month hold ≈ $21K in tax/insurance + $40K in loan interest.

Points

Loan Points

A one-time loan fee; 1 point = 1% of the loan amount. Hard-money loans commonly used for flips typically charge 1-3 points.

Example: $600K loan at 2 points → $12K paid up front.

MLS

Multiple Listing Service

The shared listing database used by US agents — this tool's data source (via the Trestle API). Quote the MLS # when talking to an agent.

Example: MLS # OC26-12345 can be sent straight to your agent.

Flip Pair

Historical Flip Pair

A buy-then-sell record of the same home (a repeat sale). The tool learns each area's real flip spreads and timelines from these.

Example: Bought $810K in Apr 2023 → sold $1.07M in Nov 2023.

Verdict Rules

These rules are rendered from the same file the verdict engine runs — they cannot drift apart.
  • ⚪ Insufficient Data: fewer than 3 comparable sales, or valuation confidence below 40/100. The model abstains from a conclusion; if you are still interested, a manual pricing review is required.
  • 🟢 Worth Viewing: the conservative case (valuing the home at the low end of its range) does not lose money, evidence is at least Moderate, the composite score is ≥ 80, and there is no high "already renovated" risk flag.
  • 🟡 Negotiate First: the deal does not pencil at list price, but works at the model's target price, or negotiation signals are strong (negotiation score ≥ 20).
  • 🔴 Not Recommended: little or no profit even at the target price, or too many risk flags.
  • A green light means the home is worth your time to view in person — it is not a buy order. Every verdict comes with its "why" and its "what could go wrong".

Underwriting Assumptions

The shared assumptions behind every profit estimate
ItemValueWhat it means
Rehab factor50% of historical per-sqft renovation cost for similar flipsAn optimistic screening assumption used only for first-pass ranking; replace it with contractor bids when you have them.
Property tax1.25% / yrAccrued at 1.25% of the purchase price per year while you hold — the longer the hold, the more you pay.
Selling costs4%4% of the sale price, covering agent commissions and seller-side closing costs.
Buyer closing1%1% of the purchase price, covering title, escrow, and other buyer-side closing costs.
Loan-to-value75%Assumes 75% of the purchase price is financed; the remaining 25% is your own cash.
Annual interest10%A typical hard-money loan rate for flips, accrued over the holding months.
Loan points2 pointsA one-time fee of 2% of the loan amount, paid at funding (1 point = 1% of the loan).
Hold periodModel-estimated per dealExpected months from purchase to resale, inferred from historical flip timelines in the area; it drives taxes and interest directly.

All profit figures are pre-tax. Non-resident status (e.g. FIRPTA withholding) can materially change your take-home — consult a tax advisor.

Model Track Record

published so you can judge how much to trust the numbers

Applying today's cost assumptions to the 8,995 historical flips the pipeline has detected (real purchase and resale prices, modeled costs):

  • 9.5% would have been profitable under the conservative cost stack
  • 19.7% under the baseline stack (median net profit $-137,904)
  • 39.1% under the optimistic stack

Actual rehab spend per flip is unknown, so these calibrate our cost assumptions — they are not audited investor P&L. The spread between the three numbers is the honest width of the model's uncertainty.

ARV accuracy: the ARV backtest has not been run on this deployment yet (scripts/backtest_arv_model.py scores past predictions against actual sale prices and publishes the error here).

Why is a home missing?

Listings matching any of the following are hard-filtered and never appear in the candidate list:

  • Not an active listing: Homes under contract, pending, or closed never enter the candidate pool; Coming Soon listings are not yet open and are not underwritten either.
  • Missing coordinates or key fields: Without coordinates, square footage, or a list price the model cannot find comps or run the numbers, so the listing is skipped.
  • Built too recently: Homes built in the last few years usually leave no renovation upside — they are not what this tool looks for.
  • Description indicates fully renovated: Listings described as fully remodeled or turnkey have most of the flip upside already priced in, so they are not treated as rehab candidates.
  • Too few comps to underwrite: When too few similar nearby sales exist, the model cannot produce a trustworthy valuation — better to show nothing than to dress up a guess as certainty.

For a case-by-case check, contact the data team.

Data Status & Refresh

Loading data status…

FAQ

Why are there so few green lights?

A green light requires no loss even in the conservative case, evidence of at least Moderate, and a top composite score — rarity is by design. The tool's value is filtering out the vast majority of listings that are not worth your time, not padding a page with good-looking numbers.

What does a score of 86 mean?

The composite score is a 0-100 blend of rules and machine-learning models, used only for ranking. It is not a profit probability: 86 does not mean an 86% chance of making money — it only means the deal ranks ahead of the vast majority of active listings.

How often does data refresh?

Data refreshes automatically once a day (see Data Status & Refresh above for the exact time), covering new listings, price cuts, status changes, and delistings. The nav bar shows when data was last updated.

Where does the data come from?

Listing data comes from the US MLS via the Trestle API. Listing text is shown as-is from the source — the tool does not rewrite it, to avoid introducing errors. Key terms are explained in the glossary above.

Why can the profit range go negative?

The conservative case assumes the home sells at the low end of its value range; a negative number means the deal can lose money if the market softens or the valuation proves optimistic. Read the left end of the range first and ask whether you could absorb that loss before deciding to view.