Same Bitcoin, two exchanges: 100,000 USDT on one venue, 100,400 on another. That 0.4% gap looks like free money. It isn't — after fees on both sides, slippage, withdrawal costs and 30 minutes of on-chain limbo, roughly 0.1% is what actually lands in your account. Cross-exchange arbitrage is real and direction-neutral, but it pays for cost accounting, not for speed-demon optimism.
👉 Read the full guide (Chinese)
Three plays, three different games
| Play | You earn | Structure | Main risk |
|---|---|---|---|
| Spread arbitrage ("brick moving") | The spatial gap between two spot books | Buy on the cheap venue, sell on the rich venue | Price moves while coins are in transit |
| Funding-rate differential | The gap between two perp funding rates | Short the high-funding venue, long the low-funding venue | The two venues' prices diverge → hedge gap |
| Triangular arbitrage | Pricing inconsistency in a 3-pair loop | USDT → BTC → ETH → USDT inside one venue | Bots eat it in seconds; humans lose |
Spatial gap, time gap, pricing gap — increasing execution difficulty in that order.
Where the 0.4% goes
| Cost item | Typical (regular user) |
|---|---|
| Spot fees ×2 | −0.20% |
| Slippage ×2 | −0.10% |
| Withdrawal fee | −1–5 USDT per move |
| Arrival delay | 10–60 min of price exposure |
| Net per round | ≈ +0.08–0.1% |
Six or seven of every ten points of gross spread go to costs. That's the whole game — and why "I saw a spread" and "I made money" are two different sentences.
IF-THEN rules for running it
- IF you move coins between venues regularly → THEN use two-sided pre-positioning: hold BTC + USDT on both venues, buy on A and sell on B in the same second. Transfer risk drops from "every round" to "rebalancing time only".
- IF you run the funding-rate differential → THEN keep both legs strictly 1:1, use low leverage with a 2–3× margin buffer, and treat "the rate gap converges to zero" as your exit signal — not "it flipped against me".
- IF you spot a spread on a top pair → THEN first check it clears fees + slippage + withdrawal cost + a ~0.1% safety pad. Below that line, the spread is an illusion.
- IF one withdrawal gets stuck → THEN freeze new rounds on the other venue immediately. A stuck leg is a naked position.
- IF you're tempted to triangular-arbitrage by hand → THEN don't. Your counterparties are colocation bots running millisecond latency. Understand the mechanism; leave the execution to them.
Bottom line
Cross-exchange arbitrage is honest, direction-neutral income — for people who cost-account every basis point before they click. The spread is not the profit. The spread minus everything is the profit.
Top comments (0)