How Our Deal Ratings Work

Two feeds fill one board. The radar tracks a curated list of graded slabs and re-checks their live eBay listings all day. The hunt sweeps eBay's live Buy It Now listings more widely — gem-mint slabs, lower-grade slabs, and raw cards with a stated condition — and prices every one it can identify. The rail at the top sorts the board by kind (Slabs are graded cards of any grade; Raw is ungraded, Near Mint or Lightly Played — the two together are the whole board) and by delivered price in four ranges that never overlap (Under $25, $25–50, $50–100, Over $100). Gems live inside Slabs as the first grade choice, “Gems · 10”: any grader's 10, CGC's Pristine 10 and a BGS 9.5 all count as one gem grade. Every find shows the same things: what it costs delivered, what we compared it to, and how far apart those two numbers are.

Every find is priced against a benchmark, and the tile says which one. For graded cards we start with sold comps for that exact grade, via Card Hedge: the median of the last five sold-day closes inside 60 days when there are at least two. When Card Hedge only reports the grade's latest price without the day it sold, we use that figure and date it “as of” the day we checked — never as a sale that closed that day. Nothing older than 60 days backs a benchmark. For raw cards we use the TCGplayer market price (Lightly Played copies at 85% of it; cards worth under $5 stay unpriced, because shipping swamps the comparison). For the slabs our radar tracks, we also compute a live ask median: every live Buy It Now asking price, shipping included, with the highest and lowest 10% thrown out (one wildly optimistic seller shouldn't move the number) and the median of what's left. Asking prices run a little higher than what cards actually sell for, so that one is a conservative line.

When more than one benchmark exists, the lower one wins — a verdict is never flattered by the softest comparison. And we don't publish a benchmark at all until there is enough data for it to mean something; a find we can't price is labeled “No comps yet” and stays out of the main grid until a later sweep can price it. A delivered price under half its benchmark is treated as implausible — usually a different variant or a mislabeled listing — and never published.

The discount on each find compares the delivered price — item plus shipping, never the sticker alone — against its benchmark. Steal means at least 20% below. Good buy means 10–20% below. At market means within 10% either way. Those three are the grid. Finds more than 10% above their benchmark, and finds with no benchmark, sit in a collapsed tail under the board with their counts — kept out of the way, never hidden.

A find is flagged as hot at 35% or more below when you'd also clear meaningful margin after selling fees if you flipped it — our “est. net if resold” figure, which assumes roughly 15% in fees. The tile's savings line is the plain dollar gap.

The 0–100 rating is a tiebreaker for your attention. The discount does most of the work, and three smaller signals round it out: the seller's track record (feedback percentage and volume), how many real photos the listing has (actual photos beat a lone stock image), and how fresh the listing is (newer listings are less picked-over). A 90 is exceptional; a 55 is merely good.

Most of the radar's value is in what it refuses to show you. We automatically drop listings from sellers with fewer than 10 feedbacks or under 95% positive; anything whose title or condition notes mention replicas, proxies, customs, cracked or empty slabs, or damage; lots and bundles; listings located outside the US (for now); and auctions (Buy It Now only — auction coverage is on the roadmap). Raw cards make the board only with a stated condition of Near Mint or Lightly Played; anything vaguer is unpriced. Graded slabs carry a cert number and grade, which is what lets us price them against sold comps for that exact grade.

On card pages and price guides, some live listings carry a quiet comparison — “12% under last sold $1,000” — measuring the delivered price (item plus shipping; no shipping stated, no comparison) against the last recorded sold price for that exact grade, via Card Hedge, or against the TCGplayer market price for ungraded copies. A listing only gets one when the benchmark is fresher than 60 days and the title names exactly one grader and grade — lots, bundles, and ambiguous titles simply show nothing, because no comparison beats a shaky one. Past 75% either way we just say “well above” or “well below”: a gap that wide usually means a different variant, not a bargain.

Our radar's scan runs every hour and works through its tracked cards in rotation, so each card is re-scanned about twice a day; the hunt sweeps eBay for new finds every hour as well, and a separate pass re-checks live deals for sold or ended listings every hour, the board's first page first. The board is never served from a cache older than a few minutes, and every deal shows when it was last checked. When a deal's listing sells or ends, we note that it found a new home and how long it lasted — then it rolls off the board within 24 hours. We'd rather show you an empty board than a stale one.

The outbound links to eBay are affiliate links: if you buy through one, eBay pays us a small commission. It never changes your price — a deal costs exactly the same with or without our link. We never sell the cards on the board, never touch the transaction, and never rank deals by what pays us (the math above is the whole ranking). We also run our own eBay store, clearly separate from the deal board.