There is no such thing as looking up what a card is worth. There is only looking up what copies of it have sold for, deciding which of those sales were about the card in your hand, and reading the middle of what survives. That is the whole job, and it takes about four minutes once you know the order to do it in.
▸ THE QUICK VERSION
Identify the exact printing first. Set, number, language, finish. Get this wrong and everything after it is noise.
Filter to sold, not for sale. An asking price is one person's opinion. A completed sale is two people agreeing.
Delete the sales that are about a different object — graded copies, lots, other printings — then take the median of what is left. Not the newest, and not the highest.
Write the date beside the number. An undated price is a rumour within a month.
Start with which card you are actually holding
FOUR MARKS, NOT ONE NAME
Almost every wildly wrong valuation starts here rather than in the arithmetic. A card name is not an identity. The same illustration gets reprinted for years, in several languages and several finishes, and those versions can differ in price by an order of magnitude while looking identical in a thumbnail.
So before you open a search box, read four things off the card itself.
Write those four down. If the set symbol is unfamiliar, the collector number still helps: the figure after the slash is the set size, and a set size narrows the candidates quickly. Doing this first is what stops you spending twenty minutes comparing your plain copy against eight sales of the reverse holo.
Ask the question in the right tense
SOLD, NOT FOR SALE
The default view on every marketplace is what is currently listed. That view answers a real question — what would it cost me to buy one today — but it is not the question you are asking, and the two answers diverge most on exactly the cards people care about.
The vocabulary differs by site, so bridge it once and stop worrying about it.
One search habit matters more than which site you pick: search the printing, not the name. Put the set, the number and the finish into the query, then read the results with those four marks in front of you.
An asking price has one signature on it. A sale has two.
Throw out the sales about a different object
MOST OF THE LIST IS NOISE
This is the step people skip, and it is where the money is. A page of sold listings for one card is not a page of comparable sales. Below is the shape of the problem: fourteen sold listings for a single card across one ninety-day window, laid out by what each one actually sold for.
Look at what that does to a careless glance. The six discarded sales are the six most eye-catching numbers on the page, and every one of them sits above every sale that counts. Somebody skimming the top of the list walks away believing this is a hundred-dollar card. It is a high-forties card that happens to have a graded population.
The disqualifiers are always the same four:
- Graded copies. A card in a slab is a different product with a different buyer. Never mix graded sales into a raw comparison, in either direction.
- Lots and bundles. Four cards for $88 is not an $88 sale. It is not a $22 sale either, because nobody has told you how the value was distributed inside it.
- A different printing. The four marks again, and by a distance the most common of the four errors.
- Stated damage. Heavily played, creased, water-marked. Good data about that condition, useless for yours.
Read the middle
MEDIAN, NOT NEWEST, NOT HIGHEST
Eight sales survive: $41, $43, $44, $46, $49, $52, $55 and $58. There are three numbers you could take from that, and only one of them is the answer.
THE NEWEST SALE
$58
One event — and the figure most people quote, because it is at the top of the page.
THE AVERAGE
$48.50
Fine here, because nothing extreme survived the filter. A single bad comp drags it.
THE MEDIAN
$47.50
The middle of the eight. What the card does on an ordinary day, which is the question.
The average and the median agree closely here, and that agreement is itself information: it says the eight sales are a tight group with no single result doing the work. When those two figures pull apart, one sale is dominating — and the median is the one that ignores it. That is the entire reason to prefer it.
Eight comparable sales in ninety days is a healthy card. Two is not a market, it is an anecdote — widen the window before you widen your confidence.
The two you cannot read
WHEN AN OFFER WAS ACCEPTED
The dashed marks are the quiet trap. When a listing sells through an accepted offer rather than at the asking price, eBay does not publish what the buyer actually paid; the figure carried on the sold listing is the seller's asking price. Take those at face value and you have quietly imported two asking prices into a set of sales, which is the exact mistake this whole chapter exists to prevent.
There are two ways to see through it. eBay's own Terapeak research tool, under the Research tab in Seller Hub, reports the transacted figure. Outside eBay, 130point surfaces the same thing from its sold-listing search. If you can get neither, do the honest thing and leave both sales out — then say your answer rests on eight.
Date the number, then stop
THE LAST TEN SECONDS
Write it as a sentence rather than a figure: plain print, raw, near mint — median of eight sales, $47.50, ninety-day window ending today. That sentence survives being read back in six weeks. A bare "$47.50" in a notes app does not, because you will not remember whether it covered the holo, whether you excluded the graded pair, or how old it already was on the day you wrote it.
It changes arguments, too. When somebody tells you the card is worth a hundred dollars, you are no longer trading opinions with them. You are asking which sales they are looking at — and most of the time the honest answer is the four you already threw away.
THE REST OF CHECKING PRICES
ELSEWHERE IN CARD VALUES