Follow the evidence behind the range.
Updated 24 October 2026. An observed range describes eligible sales returned for a selected catalog card. It is not a physical authentication, a predicted grading result or a guaranteed selling price. No OpenAI or other language model generates the displayed prices.
1. Identify and confirm the exact version
Enter the year, manufacturer, set, player and card number. Keep copy number and print run separate: 12/99 is one copy of a /99 issue. Choose the catalog card and exact Base or named parallel, then confirm it against your physical card. Search ranking and a seller’s title do not establish identity. Where the provider omits a separate Base-availability flag, the interface says so.
2. Keep condition and grading groups separate
Raw and graded records are queried separately. Graded selection requires a specific company, service and grade from the provider’s reference catalog. PSA 10 is not interchangeable with another company’s 10. Raw observations can cover different physical conditions; inspect the listing photographs yourself.
3. Retrieve a bounded sales window
CardSight supplies recent auction observations on request. You choose 30 or 90 days; the date anchor uses the provider’s US Eastern calendar. Our server explicitly requests the selected parallel and grade, with a maximum of 500 records. Base records missing a parallel field are accepted only when our server explicitly requested Base and the response echoes that Base filter. Other unknown or conflicting variants are excluded.
4. Check records without inventing missing values
We exclude fixed-price asking records, unknown listing types, mismatched variants or grading groups, missing sources, unsafe URLs, invalid prices/dates, out-of-window records and duplicate listing links. Certain lot, reprint, damage or qualifier words in titles trigger exclusion for review. This is a limited text check, not an image inspection or a guarantee that every problematic listing was detected.
High or low prices are not discarded merely because they are far from the others. Review surprising results at their source. We do not retrieve hidden accepted Best Offer amounts in this workflow.
5. Show sample size, dates and limitations
At least three eligible observations are needed for a minimum, maximum and median. Zero, one or two observations produce an insufficient-sales state; available individual records can still be inspected. USD prices are provider-reported. Shipping, taxes and buyer-premium treatment may be unconfirmed, and the figures are not net sale proceeds. A record cap or provider advisory is shown as limited coverage.
6. Keep indirect evidence outside the exact range
Other players, colors, print runs and grades can be discussed as explicitly indirect context. We do not multiply Base prices by a rarity factor, interpolate /99 from /199 and /49, or silently combine those records into an exact-match range. Our current web tool does not provide a multi-year archive or an automatic grading break-even calculator.
Requests can be unavailable because of provider errors or rate limits. The eBay search links remain a separate research option. See how these requests are processed and our correction policy.