Arbitrage is the closest thing finance has to a free lunch: buying an asset in one place and selling it in another at a higher price, capturing the difference without taking a view on which way the market will move. Foreign exchange, with its thousands of banks and dealers quoting prices around the clock, sounds like it should be full of these gaps. In theory it is — in practice, the gaps that matter close in milliseconds, and the prices a retail trader sees are shaped in a way that usually closes them before they open. Understanding why explains a lot about how the forex market actually works, and it is a useful filter when a product or service claims otherwise.

What arbitrage actually means

An arbitrage trade has three ingredients: the same instrument (or an exact equivalent) priced differently in two places, simultaneous buying and selling, and a locked-in profit that does not depend on market direction. This is different from speculation, which accepts market risk in exchange for potential return. It is also different from the carry trade, which earns a rate differential but stays fully exposed to exchange-rate moves — the carry trade is often loosely called "arbitrage" in marketing, but it can and does lose money.

Arbitrage is not just a way to profit from price gaps; it is the reason the gaps are so small. Every time a trader exploits a discrepancy, their own buying pushes the cheap price up and their selling pushes the expensive price down. Across millions of participants and modern electronic infrastructure, this feedback loop keeps prices aligned almost continuously. A 2025 survey by the Bank for International Settlements estimated average daily turnover in over-the-counter foreign exchange markets at $9.6 trillion, with the US dollar on one side of 89.2% of all trades. A market this deep, with this many participants watching the same quotes, does not stay mispriced for long.

The price you see has two sides

Before looking at cross-currency arbitrage, it helps to remember what a forex quote actually is. A platform showing EUR/USD at 1.0850 is showing two prices: the bid, at which you can sell, and the ask, at which you can buy — with the spread between them. You can never buy at the bid or sell at the ask. Any arbitrage calculation built on "the price" rather than the side you can actually transact at is built on fiction.

This matters because retail forex and CFD platforms typically quote their own prices, often derived from liquidity providers plus a markup. That markup — the spread, and any commission — is the first hurdle any arbitrage attempt must clear. Retail spreads on major pairs are often a fraction of a pip, but as we will see, genuine cross-market price discrepancies in the interbank market are frequently even smaller.

Triangular arbitrage: three currencies, one implied rate

The classic forex version is triangular arbitrage. Every currency pair's price is mathematically implied by two others through the dollar: if EUR/USD is 1.0850 and GBP/USD is 1.2690, then EUR/GBP should trade at 1.0850 ÷ 1.2690, or about 0.8549. If some dealer quotes EUR/GBP meaningfully away from that implied cross rate, a trader can cycle through three trades — dollars to euros, euros to pounds, pounds back to dollars — and, if the numbers line up, end with more dollars than they started with, whatever happens to exchange rates afterward.

A textbook illustration (from the international-finance literature, using constructed bank quotes): with EUR/USD bid at 0.8171 euros per dollar, GBP/USD bid at $1.4650 per pound, and a bank selling pounds at €1.1910 while the implied cross rate is €1.1971, a $5,000,000 cycle yields about $5,025,406 — a profit of roughly $25,406, or about 0.51% (51 basis points). The arithmetic checks out; we recomputed it with exact figures during drafting.

The catch is in the word "constructed". That example comes from an era of bank desks quoting each other by phone. Studies of modern high-frequency data tell a very different story about size and duration.

Why the windows close in milliseconds

Research examining high-frequency exchange-rate data (cited in the standard literature on triangular arbitrage) found that arbitrage opportunities between the Japanese yen and Swiss franc did appear — but 95% of them lasted five seconds or less, 60% lasted one second or less, and 94% were smaller than one basis point. One basis point means a potential profit of about $100 on $1 million transacted.

Now compare that with the retail hurdle. Using constructed but realistic retail quotes — EUR/USD 1.0850/1.0852, GBP/USD 1.2690/1.2694, EUR/GBP 0.8540/0.8548 — a $10,000 three-leg cycle (buy euros, convert to pounds, convert back to dollars) ends at about $9,986.42: a loss of roughly $13.58, or 13.6 basis points, purely from crossing the spreads. Even if the EUR/GBP quote were mispriced by a full basis point from the implied cross — larger than 94% of the opportunities in the study — the cycle still loses about 8.9 basis points. The spread alone is a bigger obstacle than the discrepancy is an opportunity. (All quotes here are constructed for illustration, not observations from any live platform.)

Duration matters too. Capturing a one-second window requires identifying the discrepancy, sending three orders, and having all three fill at the quoted prices — faster than competing participants doing the same. The literature describes an "electronic trading 'arms race'" among firms competing on speed, with infrastructure costs that retail setups cannot match. There is also a structural asymmetry: an arbitrageur who uses limit orders risks legging out — one leg fills and the others do not, leaving an open directional position that must be closed at whatever the market now offers. That cost can wipe out many small wins.

The retail reality

Bringing this together, several layers sit between a retail trader and an executable arbitrage opportunity:

• Spread and commission. As the worked example shows, ordinary retail transaction costs typically exceed the size of genuine interbank mispricings.

• Quote consistency. A single broker's pairs are usually internally consistent by construction, since the prices come from the same liquidity aggregation — the exact condition for triangular arbitrage does not arise on one screen.

• Speed. Millisecond windows are contested by dedicated firms with co-located infrastructure. A retail connection adds latency at every hop.

• Execution method. Many retail platforms use dealing models that can requote or reject orders when the underlying price has moved — precisely the protection a price discrepancy needs to survive long enough to trade.

• Terms of service. Broker agreements commonly restrict certain automated or latency-exploiting strategies, and accounts engaging in them may face order rejection or closure. Exact rules vary by broker; read the specific terms rather than assuming.

None of this makes arbitrage a myth. It is real and it disciplines the market — the studies cited above observed it directly. The point is who can actually capture it: largely banks and specialized high-frequency firms, for whom it is a technology business with thin per-trade margins and heavy fixed costs, not a strategy a standard retail account can replicate.

Practical checks

If you encounter an "arbitrage" product, a few checks help:

1. Ask which two prices are being compared. A legitimate explanation names the venues and the price basis (bid/ask, same settlement date). Vagueness here is a warning sign.

2. Check whether the claimed return survives the spread. Recompute any example with the sell side of the quote, not the mid price.

3. Separate arbitrage from other strategies. Carry trades, covered interest strategies and statistical convergence are risk-bearing trades, not riskless arbitrage, whatever the label says.

4. Read the broker's terms on automated strategies before deploying any EA marketed as exploiting price gaps.

FAQ

Is forex arbitrage illegal? No. Arbitrage is a legitimate and economically useful activity — it is what keeps prices aligned across venues. However, brokers may restrict specific practices (such as latency exploitation or abusive automation) in their terms of service, and some execution practices are separately regulated depending on jurisdiction. The legal and contractual picture is distinct: lawful in principle, still subject to the rules of your account agreement and your regulator.

If big firms profit from it, why can't I? The same economics that makes arbitrage real makes it inaccessible at retail scale: opportunities are tiny (mostly around one basis point in the cited study), fleeting (most under five seconds), and captured by firms competing on infrastructure speed. Fixed costs of that infrastructure dwarf what a small account can recover from thin margins.

Why did my broker reject or reverse my arbitrage trade? Brokers' terms commonly address automated and latency-sensitive strategies, and execution models may requote or cancel orders when prices move between request and execution. Whether a specific trade is permitted depends entirely on your broker's agreement — there is no universal rule, so check the actual terms and your jurisdiction's rules before relying on any such strategy.

Conclusion

Forex arbitrage is a real mechanism, and its effects — tight cross-rate consistency, narrow spreads, fast price correction — are visible everywhere in the modern market. The opportunities themselves, however, are small, brief, and largely captured by well-equipped institutions. For a retail trader, the useful takeaway is not how to catch the next gap, but how to read claims about it: check which prices are being compared, whether the spread is included, and what the broker's own terms say. Those checks cost nothing and filter out a lot of noise.