What these prices mean
Every figure here is built by our own pricing engine from prices and sales we record ourselves, every day. This page explains what each number means, how close it lands to reality, and where it is weak.
- Card markets
- 53,000+each printing priced separately
- Typical accuracy
- ±11.2%on markets we rate reliable
- Range holds
- 68%of sales land inside it
Three prices, never merged
A card does not have one price, and pretending otherwise hides the useful part.
Last trade
What someone actually paid.
The only figure representing money that changed hands. It is often weeks old, so we always show its age — a sale from 60 days ago means something very different from one yesterday.
Best ask
The cheapest one you can buy right now.
Priced on item plus postage, because a $40 card with $12 shipping is dearer than a $48 card posted free. An ask proves someone wants a price, not that anyone will pay it.
Mark
Our estimate of fair value, with a range.
Always shown as a range, never a lone number, and always labelled with what it was built from. If the three prices disagree, that disagreement is the most useful thing on the page.
How the mark is decided
Best available evidence wins. The label tells you which rung you are on.
Anchored to recent sales
Enough verified sales in the last few weeks to speak for themselves. The estimate leans harder on them as more accumulate.
Carried forward
The last real sale is too old to speak for today, so it is walked forward by how that card’s peer group has moved since. The range widens the older the sale gets.
From the catalog price
No sales we can verify, but a market price for that exact printing. Useful, and clearly weaker than a real sale.
Estimate only
Nothing direct. Shown as an estimate and never headlined.
The mark cannot jump more than 15% in a day unless a real trade backs the move, and it can never drift far from its own recent average. That is what stops one unusual listing repricing a card.
Each printing is its own market
The single most important correction we have made.
Same card number, nine times the price. The same is true inside a single product — a normal and a reverse holo of the same card can differ four-fold. Treating those as one price produced charts that sawtoothed between unrelated levels and a movers board where every entry was an artifact.
Every printing now has its own history and its own ticker, and a card’s headline price is its standard print — never the rarest or dearest one. Defaulting to the dearest, as we once did, overstated collections.
How accurate it is
Measured against real sales the estimate never saw, not asserted.
| Cohort | Sales tested | Typical miss | Landed in range |
|---|---|---|---|
| Every market | 6,526 | 13.5% | 68.0% |
| Markets we rate as reliable | 3,675 | 11.2% | 69.2% |
| Markets we flag as unreliable | 2,851 | 18.3% | 66.6% |
We re-price every recorded sale using only what was known before it happened, then compare. A range is only worth having if reality falls inside it about as often as it claims — ours does, 68% of the time against a 68% target.
It did not start there. The first version published a range roughly a third as wide as it should have been, and reality landed inside it only 28% of the time. We measured that, corrected it, and re-measure monthly.
What we refuse to show
A board that occasionally misses a real move is useful. One that invents moves is worse than none.
Movers
Four tests, all of which must pass.
Compared as medians over two windows rather than one reading against another; every recent reading must agree on direction; the window must be internally consistent; and penny cards are excluded, because 9¢ to 31¢ is “+244%” and means nothing.
Changes
A dash, not a zero.
When there is not enough history to state a change, we show “—”. Printing 0.0% would be a claim that the price held, which is a different thing from not knowing.
Depth
How much evidence sits behind a price.
Completed sales count most, because they are the only proof a price can be realised. Live listings count less — asks are intent. A dense price feed on a card nobody trades is not liquidity.
Sparse sales
Never smoothed into a curve.
A card may trade twice a month. Real sales are drawn as individual points, and any gap we bridge is drawn dashed. Smoothing sparse sales into a confident line reads as precision nobody has.
Where it is weak
Stated plainly, because you will find these anyway.
There is no bid side
Nowhere publishes standing buy orders for cards, so our order book shows asks only and says so. We do not mirror it into a fake symmetric ladder.
Sale coverage is uneven
No marketplace offers a complete feed of completed sales at any price. Every sale we hold was captured by us over time, so coverage is far better on popular cards than obscure ones.
No population data
No grading company exposes population figures through an open interface, and we will not scrape them. So our indices are equal-weighted: weighting by scarcity would rest on a number we cannot honestly source.
Sales are not split by printing
Listing titles rarely state which printing sold, so a sale is attributed to the card and shown with its grade rather than silently assigned to one printing.
We measure the typical miss, not the worst one
The figures above are medians. The tail is not yet published, and the tail is what matters for anything involving borrowing — so nothing here sizes a loan.
Grading changes everything
A graded sale is not evidence about a raw copy. Grades are kept on their own ladder and never folded into a card’s raw price.
Prices are informational. They are not an offer, a valuation, or investment advice.