Grailed.gg

Load gems with crypto, pick a pack, and rip it. Each pack reveals one Pokemon card from a published pool, and from there you either ship it to your door or sell it back for gems instantly.

The packs run from a $0.50 Newbie up to a $245 Haunted, with most of the range sitting between $1 and $65. Every card in the catalogue is English, and the top of the pool is genuine vintage — a Crystal Celebi, a 1st Edition Dark Charizard and a Crystal Nidoking all sit above $18,000 at Grailed's own valuations.

There's also Pack Battles, a player-versus-player mode where two or more people rip simultaneously and the bigger total takes the lot, plus a leaderboard, an affiliate scheme and a loyalty store still marked as coming soon.

This is where Grailed is unusually open. Its catalogue exposes each pack's whole pool inline: card name, set, year, the value Grailed assigns it, and the exact probability of pulling it. Nothing is bucketed or rounded into a band.

We verified the arithmetic. Across all fourteen packs the per-card probabilities sum to exactly 1.0000 — no rounding drift, no missing mass, no unlisted "everything else" tier absorbing the remainder. Plenty of sites publish odds that quietly fail that test.

Grailed also publishes its own "risk level" per pack, which is a volatility measure rather than a return one — the Newbie pack reads 83% risk, the Starter 18%. It tells you how top-heavy a pack is, not how much comes back.

Every number above is Grailed's. It chooses which cards go in a pack, sets the value on each one, and publishes the odds. None of that is independently priced, and the values are what a sell-back would be measured against.

There is a "Live Wins" rail on the site showing recent pulls, and it does carry the card, its value and its odds. But it only broadcasts rare pulls. We compared it against the published pools across seven packs and it never once showed the pack's most common card — the 30%, 29.7%, 28.9%, 24.8%, 19%, 18.9% and 18.8% cards were all absent, while cards as rare as 0.01% appeared repeatedly.

That matters more than it sounds. A feed that only shows the good outcomes cannot be used to measure what a pack really returns, and it makes the site look far luckier than it is. It's why Grailed appears here as a review and not on our measured board — we rank packs on what real pulls came back, and there's no public record of that here.

It presents itself carefully: cards are described as PSA-authenticated where graded, the site states plainly that it is independent and not affiliated with Pokemon or Nintendo, and it publishes a fairness page along with support and partnerships contacts.

The openness of the catalogue is a real point in its favour — a site intending to mislead does not usually hand over its entire pool with exact probabilities attached. The probabilities holding to exactly 1.0000 is a further small mark of care.

What is missing is outside verification: no third-party pricing, no published sell-back rate, and no complete record of pulls. Its own configuration reports a 0% sell fee and 0% withdrawal fee, which is notable if it holds, but we have not put a withdrawal through to confirm it.

Read the pool before you buy — it's right there, which is unusual, so use it. Look at where the probability actually sits rather than at the grail on the banner: in several packs three cards carry most of the weight.

Start cheap. A $0.50 or $1 pack is a real way to see how the site behaves before committing, and the low end here is lower than almost anywhere else we track.

Ship the cards you actually want. Selling back converts a real card into site currency at Grailed's own valuation, which is the point where you're most dependent on numbers nobody outside the site sets.

And treat the published EV as a claim. It's computable exactly from their data, which is more than most sites allow — but computable isn't the same as verified, and it's not what we rank on.