Pokemon Card Pricing Methodology

A card's market price is built from completed sales, not asking prices. A graded card's market price comes from trusted completed eBay sales in that exact grade. A raw near mint price is a daily market price from an outside marketplace. Live bid and ask prices from the order book are shown alongside the market price, but as their own clearly labeled numbers, never blended into it.

Not every sale counts. Before a sale is used in a price, it has to pass a confidence check: outlier sales, low-confidence sales, sales flagged as invalid, and negotiated best-offer deals are all excluded. A card's price is built only from sales that clear that bar.

Except where a value is explicitly labeled as an estimate, a graded market price is computed from completed sales, not from estimates.

Raw (ungraded) prices are anchored to a near mint (NM) baseline, the condition labeled "Raw NM" across Double Holo, and that baseline refreshes daily. It is a market price synced from an outside marketplace, so the graded steps below do not apply to it. A card in a lower condition, played corners, whitening, surface wear, is worth less than a near mint copy of the same printing.

Grading changes the picture entirely, since it replaces a visual condition estimate with a verified, sealed grade. See how much your Pokemon cards are worth to compare a specific card's raw price against its graded prices side by side.

A graded price is always tied to one specific grading company and grade, for example PSA 10, never blended across grades or averaged with a different company's scale. Here is the pipeline, in order:

1. Pull up to the 5 most recent trusted sales for that exact card, in that exact grading company and grade, sold within the last 365 days.
2. Set aside any sale that sits more than 25% off the group's middle price, where recent sales count for more in finding that middle, so the latest sales set the reference point. That outlier rejection keeps a single bad-faith or miskeyed sale from skewing the price. The newest sale is never set aside this way unless it is 5 times or more (or a fifth or less) of that middle price. One fresh sale at 5 times or more (or a fifth or less) of the typical price of the others cannot reset the price on its own. This check only runs with at least 4 sales, and it is kept only if at least 3 sales survive it. If fewer than 3 sales are within 25% of that middle and it sits 5 times or more (or a fifth or less) from the plain middle of the same sales, the check is run around the plain middle instead, and the newest sale gets no exemption. On vintage cards (Sun & Moon era and older) only unusually low sales are set aside, so a genuine high sale still counts.
3. Average what is left, weighting more recent sales more heavily. The weight halves about every 10 days, so a sale from three weeks ago counts about a fifth as much as one from today.
4. If fewer than 2 sales survive that process, it is too thin to trust on its own, so Double Holo falls back to a daily pricing snapshot for that grade and marks the result as an estimate.
5. If there is exactly one trusted sale and no snapshot either, that single sale is used as a last resort, also marked as an estimate.
6. If none of the above produce a number, Double Holo shows no price. It does not guess.

A few surfaces deliberately skip part of the pipeline above. Price alerts and seller auto-pricing do not fall back to a daily snapshot, so a card with fewer than 2 trusted sales is left out rather than estimated. The partner data feed reports the single most recent trusted sale instead of the weighted average. And the raw-condition estimates a scan gives for an ungraded card are estimates, not graded market prices.

Prefer a plainer walkthrough? Read how Pokemon card pricing works on Double Holo.

Thinly traded cards do not always have enough sales history to support a full weighted comp. When a card has fewer than 2 trusted sales in its grade, Double Holo labels the price an estimate rather than presenting it as an equal match for a well-supported, multi-sale comp.

An estimate comes from one of two places: a daily pricing snapshot for that grade, or, if even that is missing, the single trusted sale on record. Either way, the label tells you the number is thinner than a full comp, not that it is wrong.