Deal Sniper

Trust & data sources

What Deal Sniper knows, how it knows it, and what it refuses to guess. The live state of every source is measured on data quality.

Where the data comes from

eBay Browse API

Live listings: what is actually for sale right now, its asking price, format and seller.

Only an adapter that opened the individual item page may mark a listing live-verified — nothing else is rankable.

eBay Marketplace Insights

Completed sales, where our application is approved for it — the strongest form of price evidence.

Approval-gated. When it is unavailable the app says so rather than substituting estimates.

MLB Stats API

Player identity, level, organization and season statistics for the prospect universe.

MLB's own public API. Statistics refresh on a schedule; the schedule feed is separate and daily.

Card checklist provider

Card identity: product, parallel, print run and whether an autograph is pack-issued.

Checklist data is what makes 1st Bowman status VERIFIED. A seller's title alone never establishes identity.

Your own imports

Purchases you import become cost basis in the portfolio and confirmed comps in the market.

Your data stays yours; it is used to price your holdings, not resold or shared.

Confirmed vs modeled

Every number on this site declares which one it is. They never share styling, and the difference is enforced in the components themselves, not left to editorial discipline:

Confirmedpriced from verified sales of the exact card.

Model estimatean estimate — no confirmed sale of this exact card backs it.

Evidence limitedreal evidence, too little of it; the sample is always shown.

Identity unverifiedwe could not prove which card this is, so we do not price it.

Data staleobserved earlier, not re-checked recently.

Two rules follow from this and are never bent: asking prices are not comps — a listing that merely disappeared from eBay is not a sale — and unknown is not zero. A price, population or ranking we cannot establish is left blank, and the section says what was missing.

How rankings are computed

The full formula, every gate and every weight is public on the methodology page, and how we score says the same thing in plain English. In short: a card must clear two independent gates — provably underpriced against confirmed sales, and a player with a real case for reaching the majors. Scores are orderings, not probabilities; there is no backtest here, so the site never prints one.

How we make money

Some outbound marketplace links are affiliate links. Rankings, scores and fair values are computed before any link exists, and no ranking input reads commission data — the full disclosure explains the mechanics. The app never buys, bids or makes offers for you; it finds cards and links you to them.

What this tool cannot tell you

  • Whether a prospect will make the majors. Nothing here is a forecast of a player's career.
  • What a card will be worth later. Bear/base/bull figures are multiples of a measured value today, not predictions.
  • Condition from a photo. Raw-card grading outcomes are estimates with wide error, and we say so where they appear.
  • The whole market. We rank what we have scanned and verified; the coverage number on every board says how much that is.
  • Whether you should buy. Collectibles are speculative and nothing here is investment advice.

Found something wrong? The misses are published too — see our calls.