The score is the heart of Worth Radar. It's not a generic price estimate — it weighs condition, effort, and real market data into a single, explainable number. Here's what each part means.
The recommendation
Every scored listing lands in one of four buckets, by its 0–100 score:
- Pursue now (85–100) — a strong, high-confidence flip.
- Pursue (65–84) — a solid opportunity worth acting on.
- Maybe (40–64) — could work; look closer before committing.
- Ignore (0–39) — not worth your time.
Profit per hour — the headline number
A $100 profit that takes five hours isn't the same as one you grab on the way home. Worth Radar weighs your time, not just the dollar profit, so profit-per-hour is shown large and green. Expected dollar profit is the supporting figure.
The effort estimate also depends on how you'd get the item. Local pickups (KSL, Craigslist, OfferUp) include a drive — a transport cost and pickup time — because you go get them. Items from shipped/online sources (Mercari, eBay) come in the mail, so there's no pickup drive: no transport cost and no drive time count against them. That keeps a shipped flip's profit and profit-per-hour honest instead of docking it for a trip you'd never take.
Condition comes first
Seller claims aren't treated as facts. "Works great" is a claim; "easy fix" is an uncertainty signal. We analyze the photos and description for functional status, cosmetic condition, completeness, defects, and contradictions between what the seller says and what the photos show — and lower the score and confidence when they don't line up.
Sold comparables — the reality check
Where available, we pull recently sold comps (e.g. from eBay) and show a Low / Average / Median / High panel. The deal rating ("32% below sold comps") is anchored on the sold median — real market ground truth, not the AI's guess — so it stays honest even if the estimate is off.
Max buy price
For a confidently-scored listing we show the highest price you could pay and still hit your target profit — your ceiling before you negotiate. It's computed by inverting the same cost model behind the score, and it hides when there isn't enough information to compute it honestly.
Honesty by design
When there's too little evidence — no photos, a thin description, low confidence — Worth Radar returns Maybe ("can't confidently judge") rather than a fake Ignore. Free items are treated the same way: a $0 listing is floored to at least Maybe and surfaced ("worth a look") instead of being buried, since "is this free thing worth grabbing?" isn't a resale-margin question — though genuine resale value can still lift it to Pursue. And every score has a "why this score" breakdown so the number is explainable, not a black box. Over time, Calibration compares predictions to your actual results to keep estimates grounded.