Deal Sniper
Deal Sniper ResearchUpdated 2026-08-29

How to read sold comps without fooling yourself

Sold-only, delivered-price, exact-match discipline — the comp-reading rules our valuation engine enforces, written out for humans.

Key finding

A comp counts only if it actually sold, matches exactly, and is compared delivered price to delivered price.

Every card-price argument on the internet is really an argument about comps. These are the rules we hold our own valuation engine to; they are just as useful done by hand on an eBay sold search.

1 — Asking prices are never comps

An active listing is an opinion; a sale is a fact. The ask you are judging must never feed the value you judge it against, or every listing looks fairly priced. Filter to sold items and ignore what unsold copies are listed at — a card “listed at $200” that never sells is evidence of the opposite.

2 — Compare delivered prices

A $20 card with $9 shipping and a $27 card with free shipping are within a dollar of each other. Add shipping to every comp and to the listing you are judging before comparing. On small Bowman cards shipping is routinely 20–40% of the trade and reverses conclusions.

3 — Exact means exact

A comp matches when player, product, card, parallel and grade all match. A Blue Wave /150 sale does not price a True Blue /150. A PSA 10 sale does not price a raw copy, and a PSA 10 and an SGC 10 are different markets. When you can't find exact matches, say so and treat any estimate as low confidence — don't quietly borrow a neighbour's sales.

4 — Window honestly, weight recent sales more

Prospect prices move on promotions, call-ups and slumps, so a six-month-old sale describes a different player. Prefer the last 30 days; if there are fewer than three sales, widen to 90, then a year — and say which window you used. Inside a window, recent sales should count more (our engine halves a comp's weight every 45 days) and the summary statistic should be a median, not a mean — one shill bid or one panic dump should not move fair value.

5 — The Best Offer discount is invisible

A large share of card sales close via Best Offer, and most price views show the original ask, struck through or not, rather than the accepted amount. Where accepted-offer amounts are visible, they routinely settle meaningfully below ask. Two consequences: ask-anchored price guides overstate value, and a “sold” price that looks strong may have been a much weaker negotiated number. Prefer sources that show true accepted amounts, and discount ask-anchored data accordingly.

6 — Watch for junk in the pool

  • Lots. “(x4)” or “lot of 5” sales are not single-card prices.
  • Duplicate international records. Aggregated sale feeds can carry the same transaction twice in two currencies; dedupe before counting.
  • Aftermarket autos. An in-person signature sold as an “auto” is a different (much cheaper) item — see the certified vs aftermarket guide.

How we apply this

Every fair value on this site is a recency-weighted median of confirmed sales, windowed 30→90→365 days with the window disclosed, exact-match only, delivered prices, with modeled estimates forced to display as low confidence. When there are no comps, we show nothing rather than a guess — a blank is honest; an invented number is not. Details on the methodology page, and you can run that exact calculation over your own list of sales in the fair value calculator.

Basis & limits

What this is built on. The rules and figures this project's own identity and valuation engines enforce, plus the domain research behind them.

Where it stops. This is an explainer, not a study: it carries no sample size and makes no forecast. Figures that move in the real world — grading fees, print runs, marketplace behaviour — can date it; the updated line above marks the last material revision.

Methods are documented on the methodology page; sources and their limits on trust & data sources.

More on what it is worth, and how we know

All what it is worth, and how we know guides →

Keep reading

See this applied to real prospects: every player page shows live listings, sold evidence and the parallel ladder for one player, and the market board ranks what is mispriced right now.