Arbitrage is the oldest idea in trading: buy something cheap in one place and sell it dearer in another at the same moment, pocketing the difference with no market risk. In forex that means the same currency value being priced two ways — on two brokers, or across three pairs that do not quite agree — and closing the gap before it closes itself. In theory it is free money. In practice the market spends every second making sure it is not.
What it is
A true arbitrage is a near risk-free profit from a price that is briefly wrong. You are not predicting anything; you are exploiting a disagreement that has to resolve. In forex those disagreements come in a few shapes: the same pair quoted differently on two brokers, or a mismatch between three pairs that are mathematically linked.
The forms retail traders usually mean:
- Two-broker arbitrage — Broker A quotes a pair a shade higher than Broker B, so you buy on B, sell on A and bank the gap.
- Triangular arbitrage — three pairs drift out of line with each other, and a loop of trades locks in the discrepancy.
Either way there is no view on direction. The profit is the gap itself, captured and closed.
How it works
Triangular arbitrage is the clearest to show. Three pairs are tied together by simple maths: EUR/JPY should equal EUR/USD multiplied by USD/JPY. When the quoted price steps out of line with the implied one, a loop of three trades captures the difference.
EUR/JPY (implied) = EUR/USD × USD/JPY
Say EUR/USD is 1.1000 and USD/JPY is 150.00, so the implied EUR/JPY is 165.00. If a broker actually quotes EUR/JPY at 165.20, you sell it and buy it back synthetically through the other two pairs, capturing 0.20, around 20 pips, in theory. Close all three legs and the currency exposure cancels out.
A worked example
On paper it reads well. You spot that 20-pip gap, and on one standard lot it is worth roughly £120 of edge. Then reality adds its charges: the spread across three pairs costs perhaps 12 pips, slippage on filling all three legs another 5, and by the time the third order lands the gap has shrunk to 3 pips anyway. Twenty pips of theory becomes a few pounds of practice, if the gap is still there at all. That is the whole story of arbitrage — the edge is real and tiny, and everything conspires to take it back.
Why it is hard for retail
Arbitrage rewards whoever gets there first, and retail is last in the queue. Banks and high-frequency firms sit microseconds from the pricing engines, while a retail order travels from a VPS through a broker to the market and back. By the time your loop fills, the gap has gone. Worse, brokers dislike arbitrageurs, because the profit comes straight out of the broker's book, so they widen spreads, add execution delays, requote, or simply close the account. What looks like free money is a race you are built to lose.
Common misconceptions
- "Guaranteed profit" ignores execution. The gap is real; filling all the legs before it closes is the hard part, and usually you cannot.
- Brokers actively fight it. Arbitrage EAs meet slippage, requotes, delays and closures. A strategy the broker can switch off is not an edge you own.
- Backtests lie badly here. Historical data cannot reproduce the millisecond fills arbitrage lives or dies on, so a perfect backtest means nothing.
- If it is sold to you, ask why. A genuine, scalable arbitrage would be run quietly by whoever found it, not marketed to retail for a monthly fee.
The Karnek note
If you run an arbitrage EA, the live account is where you learn whether the theory survived contact with your broker. Karnek reads that account from your terminal — fills, slippage, drawdown — so you see what actually happened rather than what the backtest imagined. It is read-only and can never place or change a trade.
Written and reviewed by the Karnek Research team. Last updated August 2026.
Educational content only - not financial advice. Past performance does not predict future results. Trading carries significant risk.