How CardTrends prices cards
Written for collectors. Every number on a card page traces back to one of the rules below.
One price per condition, from a fixed chain
Every card printing gets three prices: Raw (Near Mint), PSA 9, and PSA 10. Each takes the first source in a fixed chain that has a value. Nothing is blended and nothing is modeled.
- Raw (Near Mint): the PokemonPriceTracker Near Mint quote for that printing, then PriceCharting's ungraded price, otherwise N/A. "Raw" on CardTrends always means Near Mint, not a mix of conditions.
- PSA 9 and PSA 10: the median of the latest five PSA sold comps within 365 days, then PriceCharting's Grade 9 or PSA 10 figure, otherwise N/A.
A sold comp only counts when its listing title names PSA and the exact grade, matches the card, number, and printing, and describes no defect. Lots, proxies, and foreign-language cards are excluded. If two prices on a card contradict each other (a PSA 9 worth more than the PSA 10, say), the one with weaker evidence is withheld rather than shown.
Evidence strength on every price
Each price carries a label describing the evidence behind it, never a model's opinion:
- Strong: three or more PSA sales with the latest inside 90 days, or a Near Mint quote updated within 14 days.
- Moderate: plenty of sales but none recent, a few recent sales, or a provider figure updated within 60 days.
- Thin: anything older or sparser, and every PSA 9 figure that blends graders.
Priced high, in line, or low vs. its own sales
Where CardTrends has a card's completed sales history, the card page compares today's price with the median of that card-grade's sales over the previous 12 months. The price is "in line" unless it sits more than 15% away, or further when the card's own sales usually scatter more. A card needs at least 10 sales in the window to get a label.
This describes where a price sits. It is not a forecast and not a recommendation to buy or sell.
Track record
Because the band is a label, we check what happened next. At quarter-ends since the history began, every card-grade with enough sales was labeled using only sales known at that time, then measured over the following 6 months against every other labeled card at the same date.
| Label at the time | Cases | Median move vs. other cards | Share that rose |
|---|---|---|---|
| In line with its sales | 299 | +0.0% | 87% |
| Priced high vs. its sales | 146 | -7.0% | 83% |
| Priced low vs. its sales | 9 | -4.7% | 89% |
Read small samples with caution: Priced low vs. its sales has fewer than 30 cases, too few to draw a conclusion. We publish these results whether or not they flatter the label, and they update as history grows.
The CardTrends Index
The index starts at 100 in January 2023 and moves only on chained same-card returns: each month, every card-grade that sold in both that month and the month before contributes its change in median sale price, each change is capped at halving or doubling, and the index moves by their geometric average. Adding cards changes what the index can see, never its past level. Months with too few cards carry the prior level and are excluded from reported changes.
Latest level 313.9 (2026-08), +7.3% over 3 months and +111.2% over 12 months, across 2,970 card-grades.
Coverage today, and what's next
Sales history, the band, and the index currently cover WOTC-era sets (1999–2003) in PSA 9 and PSA 10, from identity-matched PriceCharting completed sales through August 31, 2026. CardTrends now keeps every qualifying sale it reads, so coverage extends to modern sets as history accumulates. See the roadmap.