How cross-border card arbitrage actually works

A Special Art Rare Charizard is the same piece of cardboard in Tokyo and in Dallas. It does not trade at the same price. That difference — the same card, cheaper in one market than another — is the entire game. Buy where it's soft, sell where it's dear, and pocket what's left after the costs. The trick is that last clause: after the costs. Most "obvious" spreads disappear the moment you subtract fees, shipping and currency conversion. This is how to tell a real gap from a fake one.

Why the gap exists

Cross-border spreads aren't a glitch. They're structural, and three forces keep them open.

• Supply. Japan is the release epicenter. Japanese-language sets, promos and Special Art cards surface there first and in volume. More local supply means softer local prices — the same card that's scarce in the US is a shelf item in Osaka.
• Collector base and demand. The US has a deep, liquid market that pays up for marquee chase cards and graded slabs. Demand pools where the buyers and the graders are, and price follows demand.
• FX and information friction. A weak yen makes yen-priced cards cheaper in dollar terms. And because Japan, China and Southeast Asia sold data has no clean public API, US buyers usually can't even see the cheaper Japanese comps. The information gap is what holds the price gap open.

A worked example

Take that Special Art Charizard. You find one landed from Japan at roughly $297. The recent sold median on TCGplayer in the US is about $384. Headline gap: $87. Feels like free money. Now run it to the end.

• Buy — landed from Japan (card + forwarding + FX): $297
• Sell — TCGplayer US sold median: $384
• Gross spread: +$87
• Marketplace + payment fees (~13% of the sale): −$50
• Outbound shipping + supplies (US domestic): −$3
• Net profit: ≈ $34

So the $87 "gap" is really about a mid-$30s net once the marketplace takes its ~13% and you cover shipping. That's still a clean margin on a single card — but it's less than half the number a raw price comparison screamed at you. Chase the $87 and you'll overpay; price the $34 and you'll actually make money.

Illustrative example only — figures are rounded for teaching. Your numbers will vary by card, condition, timing, marketplace and exchange rate.

The costs that eat the gap

Every one of these is the difference between a spread that looks good and one that pays. Model them before you buy, not after.

• Marketplace + payment fees (~10–13%). The single biggest bite. On a $384 sale that's roughly $40–50 gone before anything else.
• Shipping, both ways. Japan domestic to your forwarder, then international to you, then US domestic to your buyer. Three legs, each with a cost.
• FX and conversion spread. Card networks and forwarders skim roughly 1–3% converting yen to dollars. Small per card, real at volume.
• Condition and authenticity risk. "Near Mint" described in Japanese can grade lower in hand; centering and edges vary. Build in a haircut for the copies that come back worse than listed.
• Customs and duties. Usually minor on a single low-value card, but a line item once you're moving size.
• Time and capital. Days-to-weeks in transit is money you can't redeploy. Price the wait like any other cost.

Where to source

The buy side lives on Japan's consumer marketplaces. The ones worth knowing:

• Mercari (Japan) — the deepest consumer supply; where most singles surface first.
• Snkrdunk — authentication-forward, popular for higher-end pieces.
• PayPay Flea — another large consumer-to-consumer pool.
• Cardrush — a dedicated card retailer with structured, searchable inventory.

The catch: most need a proxy or forwarder, they're in Japanese, and — critically — sold-price visibility is thin. You can see what people are asking. Seeing what cards actually clear for is the hard part, and it's exactly where a raw browse leaves you guessing.

How Hoshino nets it for you

We built the tool to kill the guesswork and do the arithmetic you just watched — automatically, on every card.

• Scan the card. Camera or photo resolves the exact set and number.
• Pull real sold medians in each region — never asking prices.
• Net the spread. We subtract the ~10–13% marketplace fee, estimated shipping and live FX, so you see the true gap, not the headline one.
• Watch the board. The biggest-gaps board updates continuously — the cards that clear a profit today, ranked.
• Set an alert on a spread threshold and let the gap come to you.

"Buy in Japan, sell in the US" is the obvious part. The work is in the netting — and in seeing Japanese sold data at all. That's the part we do that a US-only tracker structurally can't.