How all of this actually works.

Two characters is enough to start. Searching runs as you type and every request cancels the one before it, so the table always matches the box rather than whichever request happened to come back last.

Magic comes from Scryfall, Pokémon from pokemontcg.io, and Cardsmiths Currency from a catalogue built here by hand — it has no price API anywhere. All three are merged and sorted together, and any one can be switched off.

Open a row and it shows what copies actually changed hands for — the median, the spread, and how many sales that rests on. That number is bought a card at a time from a metered source, so it is cached and shared once anybody has asked for it.

When the sales do not look like the same card — a bulk lot titled with one card's name, a graded copy in with the raw ones — the tool says the sample is too mixed to trust rather than printing a median anyway.

Cheapest, dearest or alphabetical. The "under the fee" filter keeps only cards worth less than the $80 it costs to grade one today — a raw card that would need its grade to add more than that just to break even.

A card with no price sorts last in every direction. An unpriced card is not "cheapest" — it is unknown, and floating it to the top would mislead.

Each row links out to the marketplace listing for that specific printing, where the picture and the live listings are.

Results are a window of at most 60 matches per source, so "cheapest" means cheapest of what was fetched. The footer says so whenever the set was truncated.

The long number printed on the label. It is a serial number for one piece of plastic, which is why this lookup can answer a question no card-level tool can: not what cards like this are worth, but what this one has done.

Every time it was vaulted, listed or sold on either chain, with the price and the date. Two copies of one card can trade months and hundreds of dollars apart, and this is the only place that difference is visible.

Coverage is partial. Only slabs that have been vaulted and had their label re-read carry a record, so "nothing found" means we hold nothing on it — not that the grader does not know it.

The median of other copies at the same grade, set against what a raw copy sells for. The difference is the slab premium — what the plastic is actually contributing today.

The premium is compared with the fee twice: at what you paid — the panel starts at $25, and you can type your own — and at today's $80. A slab that broke even at the old fee and would not at the new one is the ordinary case, and the honest sentence is that the waterline rose, not that somebody overpaid.

Nothing here is a valuation. Every figure is a price a copy was listed or sold at on a date; the verdict is arithmetic over those prices and the fee you tell it.

A slab can be saved beside your cards. It stays in this browser, gets re-priced on every visit, and the report on your saved list counts it — how many are above water at each fee.

Any row in a search has a save. Saved cards live in this browser with no account, and the dashboard re-prices them on every visit — a saved price is a snapshot and is never shown as today's.

Linking an email lets the list follow you to other devices and gives the nightly check somewhere to send news. The app pairs with a short code shown on screen — nothing is emailed for that. Either way, the list of cards you saved is the one thing that leaves your device, and the page says so before it does.

The address is stored by the mail provider, never in this site's database, which holds only a one-way hash of it. That is the precise claim: this database never becomes a list of email addresses.

Every card on a linked list is re-read overnight. A new printing, a card leaving Standard or getting banned somewhere people play, or a price that moved more than a little over the week — those are the events. News is not one of them, because almost nothing this cheap has ever been written about.

The first sighting of a card never alerts. With no baseline every printing would look new, and the feature would launch by sending hundreds of false notices to the people who just paid for it.

A digest goes out only when something moved. Most days nothing did and nothing is sent — a daily message whether or not anything happened is the fastest route to being filtered, and then the one that mattered is filtered too.

Subscribers can also ask for fresh sold comps on any card that has not been looked up yet. The answer is cached and shown to everybody afterwards — the subscription buys the right to ask a new question, never access to old answers.

Every row in a search has a save; a slab can be saved from the cert lookup. Everything below acts on that list, so there is nothing to notify you about until it has something in it. Saved cards live in this browser until you link them.

Look up a card →

The subscription is $2 a month. It turns on the reprint and format alerts on your dashboard — a new printing of a card you saved, or a card leaving Standard or getting banned somewhere people play — and lets you request fresh sold prices on a card nobody has looked up. Checkout links this browser with a cookie; on another device, open the billing portal from the subscription page and it finds you by the email you paid with.

The subscription →

On the dashboard, under your saved cards: "Use your cards on another device". Type an address, send the link, and follow it within the hour — nothing is stored until you do. That creates a copy of your list on the server, which is what lets the nightly check run while no browser is open, and it turns on the weekly summary of price movement by email.

This is the one thing that leaves your device: the list of cards you saved. The address itself is held by the mail provider, never in this site's database. Every email carries a link that stops the summary or deletes the list outright.

Your cards →

Under the alerts on the dashboard there is a card that fits the device you are on: Chrome and Edge offer a real "Install", Safari on a Mac does File → Add to Dock, and on an iPhone or iPad it is Share → Add to Home Screen. On an iPhone this step is not optional — a website in a Safari tab cannot receive notifications at all; one opened from a Home Screen icon can. Once installed, a notification that arrives while the app is closed leaves a mark on the icon.

The install card →

Once linked, the same panel offers "Turn on notifications". The browser asks for permission once — say yes — and then press "Send a test". A green toggle proves nothing; a notification actually appearing does. With the tab in front, it arrives as a note in the corner of the page; otherwise as a system notification. This is per browser, per device: a laptop that is on does not switch on a phone that is not.

If you have ever blocked notifications from this site, the browser will not ask again and the toggle cannot undo that. The only way back is the browser's own site settings, behind the icon at the left of the address bar.

Once linked, "Show a pairing code" on the dashboard gives a short code. Type it into the Saved tab of the iPhone app and the two lists become one — a card saved on either turns up on the other. The code works once and dies in ten minutes; it is shown on screen and never emailed. The app is being rebuilt and is not downloadable right now; the code will work the day it is.

The iPhone app →

Prices are recorded every night and printings and legality re-read shortly after. The summary goes out in the morning, US time, and only on a day something moved more than a little over the week — most days nothing did, and nothing is sent. Alerts appear on the dashboard as they are found. The first time a card is seen never fires anything.

Silence is the normal result. A message every day whether or not anything happened is the fastest way to be filtered, and then the one that mattered is filtered too.

The page starts by pricing the card and adding up what grading ($80 today), insured postage and handling actually cost, then says what the graded card would have to come back worth. For most cards worth less than the fee the answer is don't, and it is said first.

It never predicts a grade. The gap between a 9 and a 10 is usually larger than every fee combined, and a tool that quietly assumed "probably a 9" would be manufacturing the confidence this site exists to refuse.

Into Courtyard on Polygon. Out of it — redeeming the physical card. Into Collector Crypt on Solana. Out of that. A card moving between chains is two of these back to back, with a physical leg in the middle.

Courtyard has paused submissions to its vault (seen 2026-09-11), so "into Polygon" is not on offer until it reopens. Redeeming from it still works.

A rough list, how many cards, whether they are already graded, and which direction. Nothing is booked and nothing is charged; no address is asked for until something is actually being posted.

Handling is a published flat fee. Postage is at cost. Grading is paid to the grader. And the vault's own charge is stated per trip: nothing to go into either vault, $7.69 to come out of Courtyard (our own receipt), $5.99 plus $3 an extra card to come out of Collector Crypt (their published rate, more abroad). You still get a written quote for your cards — and we say plainly if the whole thing is not worth doing.

The Collector Crypt figures are their docs' word, not yet a receipt; the bridge journey is two of our own cards turning them into one. Some write-ups cite a further 2% on Collector Crypt withdrawals that is not in the docs — if it turns out to exist, it goes in the quote.

The journey, in public →

A card handled on your behalf is vaulted under the venue's own terms, and it lands somewhere you can see: our Collector Crypt account and our Courtyard profile are both public.

Collector Crypt account →Courtyard profile →

Courtyard settles completed sales on chain and keeps its order book off it, so its number is a price PAID. Collector Crypt lists on chain, so its number is a price ASKED. Every label on the page says which, because a reader who assumes ask-versus-ask draws exactly the wrong conclusion.

Same year, same number, same name, same grade — matched on the identifying fields, never on the name alone. Where the two sides disagree about the set, the pair is dropped rather than shown.

Rows with an ask far above the venue's own insured value, or a sale far below it, are hidden by default. Those are usually a mismatched card or a lot sold as one, not a bargain.

Across the whole dataset asks run about one and a half times a card's insured value while sales clear at about four fifths of it. Most of what looks like a spread is the difference between hoping and paying.

A card bought on one chain cannot simply appear on the other. It has to be redeemed, shipped insured, and re-vaulted by a different custodian — days or weeks, with a fee at each step. The tokenize page now quotes those steps, one of them from our own receipt, and usually says don't.

What the trip costs →

Two of our own cards are making that trip in public — one bought as a Polygon token, one bought physically — with every cost written down as it is paid, including the ones we cannot price yet, which are marked rather than estimated.

The journey →

Each slot is one specific sealed pack, photographed before it was listed, with what it cost written beside it. Slots are taken in purchase order — there is no shuffle and no draw. You are buying a pack, not a chance.

Claiming asks for one way to reach you — an email or a handle — and whether you want to be named on the reveal (the default is anonymous). Then it signs a line of text with your wallet: it names the slot and the price, costs nothing and moves nothing. The page then shows the SOL amount and the one address to send it to, and watches the chain for your transfer. A person confirms it landed before anything is opened.

The address is published on the slot after you claim, and nowhere else. Any address that arrives in a message, a reply or a DM is not us.

One continuous take, wrapper in frame throughout, the slot number shown before it comes off — so you can see it was your pack. The video is published after the rip and linked from the slot.

Everything in the pack is posted to you — we ask for an address only once it is opened, and postage inside the US is on us. Or, if you would rather not receive a pile of commons, we run the pull through the comps pipeline and publish what each card actually sold for.

The packs cost more than $5 and the postage is on top. That is deliberate and it is why a season has a fixed size: when the packs are gone, it stops and gets looked at again.