Crypto
Crypto Triangular Arbitrage
Triangular arbitrage cycles capital through three trading pairs, USDT to BTC to ETH and back to USDT, hunting for the rare moment the three rates don't line up cleanly enough to leave a profit after fees.
Watch the loop, execute when the edge looks real
Three rates jitter live: BTC/USDT, ETH/USDT, and the direct ETH/BTC cross rate. The cross rate doesn't always match what the other two imply, and that gap is the arbitrage. The edge readout below recomputes the full round trip, USDT to BTC to ETH to USDT, including a 0.10% fee on each of the three legs, live every tick. Click execute whenever it looks good, the tool locks that instant's rates and logs whether it actually cleared costs.
How it works
- Three pairs, one loop. Start with USDT, buy BTC, convert that BTC to ETH at the direct cross rate, sell the ETH back for USDT. If the three rates were always perfectly consistent, the loop would return exactly what you started with, minus fees.
- The edge comes from the cross rate drifting from what's implied. BTC/USDT and ETH/USDT set an implied ETH/BTC rate mathematically. When the actual quoted ETH/BTC rate on the exchange drifts even slightly away from that implied number, the loop can return more or less than it started with.
- Fees apply on every leg, three times. At 0.10% per leg, three legs compound to roughly a 0.30% drag on the round trip. The edge has to clear that drag, not merely land above zero on paper, before the loop actually makes money.
- Execution locks in that instant's numbers. Clicking execute doesn't wait for a better price, it takes the rates on screen right now, same as a real arbitrage bot would, because by the time a human decides to act the opportunity may already be gone.
Where this breaks
The edge is usually smaller than the fees, and bots get there first
Run this simulator for a few minutes and watch how often the net edge sits negative, that's realistic. Real triangular arbitrage opportunities on liquid pairs are typically fractions of a percent and close within milliseconds, because automated bots watching the same three rates execute the moment the math turns favorable, before a human can read the numbers, decide, and click. What looks like a live opportunity by the time a person notices it has usually already been arbitraged away by faster capital. The edges that remain reachable by manual traders tend to be on thinner, less liquid pairs, where the wider spreads and slippage on entry and exit can eat the opportunity as fast as the fees do.