Currency Pair Correlation and the Retail "Arbitrage" Myth
A trader I know once boasted that he had "diversified his portfolio": he had bought EUR/USD and GBP/USD at the same time, because after all they are two different pairs. A week later the dollar suddenly strengthened, both positions fell together, and his loss turned out to be twice what he expected. That was not bad luck or a broker conspiracy. It was correlation — the quiet force that makes two seemingly separate bets turn out to be one and the same. In this article I explain how to read it, why it changes, and why "arbitrage" for a retail trader is largely a myth.
What currency pair correlation actually is
Correlation is a number that describes how closely two currency pairs move to the same rhythm. It ranges from minus one to plus one. A coefficient of plus one would mean the pairs always move in the same direction, in perfect agreement. Minus one would mean a perfectly mirrored move — when one rises, the other falls by the same amount. Zero would say their movements have nothing to do with each other. In practice you will never meet the extreme values; the market operates somewhere in between, and the most interesting things happen at high but imperfect relationships.
The key is to understand that correlation is not some mystical property of pairs, but a simple consequence of which currencies make them up. If two pairs share the same currency, their fates must be linked. That is why the US dollar, present in the vast majority of turnover on the foreign exchange market, is the most common shared denominator and the main cause of the correlations we observe every day.
Positive and negative correlation in concrete examples
Take the two best known relationships. EUR/USD and GBP/USD are usually strongly positively correlated, often around plus 0.9. The reason is simple: in both pairs the dollar sits on the same side, as the quote currency. When the dollar weakens, both pairs rise; when it strengthens, both fall. From a risk point of view, buying both at once is not two trades but a single enlarged bet on the direction of the dollar. I describe the character of EUR/USD separately, because that pair usually sets the tone for the whole dollar puzzle.
The opposite pole is EUR/USD and USD/CHF, which move in mirror image, with a coefficient often around minus 0.9. Here the dollar is the quote currency in EUR/USD and the base currency in USD/CHF, so the same dollar move pushes one pair up and the other down. Add to that the closeness of the Swiss and euro-area economies. Why the franc plays the role of a safe haven and how that shapes its behaviour, I develop in the article on the Swiss franc.
"All markets are interrelated — financial and commodity, domestic and international. No market moves in isolation." — John J. Murphy, Intermarket Analysis: Profiting from Global Market Relationships, Wiley, 2004.
How correlation quietly doubles your risk
Here we reach the heart of the matter, the part that costs retail traders the most money. When you open two highly positively correlated positions in the same direction, you are not spreading risk — you are multiplying it. Suppose, purely hypothetically, that you have an account worth 10,000 euros and you buy EUR/USD and GBP/USD, risking one percent of your capital on each pair. It looks as if you have two independent trades, each risking one hundred euros. In reality, since both pairs react almost identically to the dollar, a sudden dollar rally means you lose on both at once. Your real risk to a single factor, the dollar, is not one hundred but two hundred euros — the same as a single position of double the size.
The mirror image of this problem works in your favour, if you use it deliberately. By buying EUR/USD and at the same time buying USD/CHF, two strongly negatively correlated pairs, you largely neutralise your exposure to the direction of the dollar itself. What remains is a bet on the relationship of the euro to the franc. Such hedging can be useful, but it requires understanding that you are combining two instruments into one. Solid risk management basics are a precondition here, not an add-on.
Why correlation is not constant
The biggest mistake is treating a correlation, once read, as a physical constant. The coefficient is calculated from the last few dozen observations and drifts along with whatever is driving the market. When everyone's attention is governed by one central bank's policy, dollar pairs hold together. But when a local theme comes to the fore, such as an election or one country's inflation data, a pair that was in step with others starts marching to its own beat.
Correlations change most violently in crises. In a panic, capital flees to currencies regarded as safe havens, and then everything risky falls together, regardless of historical relationships. People sometimes say that "in a crisis correlations go to one", and precisely when diversification is needed most, it stops working. The Swiss franc gave a dramatic example of this in January 2015, when the Swiss National Bank suddenly removed its cap and the mirror relationship with the euro fell apart for a moment entirely. For the wider picture of how markets move together, see this deeper guide to intermarket analysis.
The truth about "arbitrage" for retail
The word "arbitrage" sounds like a promise of free money, which is why it needs bringing down to earth. Classic triangular arbitrage relies on the fact that the EUR/GBP rate calculated by dividing EUR/USD by GBP/USD differs for a split second from the directly quoted EUR/GBP rate. In theory you could profit from that gap. In practice such discrepancies are microscopic, last microseconds, and vanish before a retail order can even reach the broker server.
They are captured by high-frequency trading systems sitting physically next to exchange servers, with latency measured in millionths of a second. A retail trader operates with latency of hundreds of milliseconds and pays the spread on each of the three pairs. It is a losing game by definition, at the level of infrastructure rather than skill. That is why the value of correlation for us lies elsewhere: in the deliberate selection of pairs, in avoiding the hidden doubling of risk, and in thoughtful hedging. It is not a free-money machine, but a tool for managing more wisely what you already have.
What to do tomorrow
- List every pair you currently hold open and mark, for each one, which side the dollar sits on — if several positions are essentially the same bet on the dollar, add up the combined exposure before you decide you are diversified.
- Open a free correlation table at your broker or on an analytics service and check the real coefficients for the pairs you trade — compare them with the rules of thumb from this article and notice where the market diverges from what you expected.
- Before you open a second position on the same day, ask yourself one question: is this a new, independent bet, or merely a repeat of the previous one — if the correlation exceeds plus 0.7, consider cutting the size of both positions in half.
- Schedule a weekly refresh of your correlations as a fixed point in your routine, and after every major central bank decision or inflation release check the table immediately, because that is exactly when historical relationships tend to break down.
- Drop the idea of arbitrage as a source of easy profit and treat correlation as a tool for defending your capital — write into your trading plan a simple rule for maximum combined exposure to a single currency, and stick to it consistently.
Currency pair correlation is one of those concepts that sound academic but in reality decide whether your portfolio truly spreads risk or merely pretends to. Understanding that two pairs sharing the same dollar are often a single bet is worth more than many a complicated strategy. Leave true arbitrage to the machines — what is left for you is something more valuable: conscious control over what you are really betting on.
Sources & bibliography
-
BIS OTC foreign exchange turnover in April 2022 · Triennial Central Bank Survey — struktura i skala rynku walutowego (7,5 bln USD dziennego obrotu) www.bis.org ↗
-
BIS High-frequency trading in the foreign exchange market · raport o handlu wysokich częstotliwości — dlaczego arbitraż w FX należy do algorytmów, a nie detalistów www.bis.org ↗
-
EBC Euro foreign exchange reference rates · oficjalne dzienne kursy referencyjne euro — wiarygodne dane do liczenia korelacji par www.ecb.europa.eu ↗
Frequently asked
Why do EUR/USD and USD/CHF move in opposite directions?
The answer is mechanical, not magical: it comes down to where the dollar sits in each pair. In EUR/USD the US dollar is the quote currency, so it sits in the denominator — when the dollar strengthens, EUR/USD falls. In USD/CHF the dollar is the base currency, sitting in the numerator — when that same dollar strengthens, USD/CHF rises. So one move in the dollar pushes one pair down and the other up, which produces a typically strong negative correlation, often around minus 0.9. The second reason is that the Swiss franc and the euro represent economies tightly linked by trade and close geographically, which is why the Swiss National Bank long worked to stop the franc drifting too far from the euro. But keep one thing in mind: this correlation is strong, yet neither perfect nor permanent. In January 2015, when the SNB suddenly removed its franc cap, the mirror relationship fell apart for a moment entirely. Treat minus 0.9 as a useful rule of thumb, not a law of physics.
Is opening two correlated pairs a form of diversification?
Usually not — and this is the trap many beginners fall into. Imagine you buy EUR/USD and GBP/USD at the same time because they look like two separate opportunities. The problem is that both pairs have the dollar on the same side and historically move together with a coefficient around plus 0.9. That means you have really placed a single bet: you are betting on dollar weakness, only at double the size. If the dollar unexpectedly strengthens, both positions lose at once, and your loss is twice what you expected from "two different trades". Real diversification means exposure to independent risk factors, not the same factor seen under two names. So what should you do? First, add up your exposure to each currency separately before you click buy. Second, if both signals are genuinely good, cut the size of each position in half so that your combined dollar risk equals that of a single trade. Third, deliberately reach for low-correlation pairs when you actually want to spread risk. I unpack this further in the article on currency pair correlations in practice.
Can a retail trader profit from triangular arbitrage?
In practice no, and it is worth saying that plainly instead of selling illusions. Triangular arbitrage relies on the fact that the EUR/GBP rate implied by dividing EUR/USD by GBP/USD sometimes differs for a split second from the directly quoted EUR/GBP rate. In theory you could buy the undervalued version and sell the overvalued one, pocketing the difference. The catch is that such gaps are tiny, last microseconds and vanish before a retail order even reaches the broker server. They are captured by specialised high-frequency systems sitting physically next to exchange servers, with latency measured in millionths of a second. A retail trader operates with latency of hundreds of milliseconds, pays the spread on each of the three pairs and commissions that eat any hypothetical profit many times over. The Bank for International Settlements, in its report on high-frequency trading, states plainly that it is these participants who arbitrage away small pricing inefficiencies in the FX market. The honest conclusion: for us, correlation makes sense as a tool for risk management and hedging, not as supposed free arbitrage.
How often do currency pair correlations change?
Constantly, though at varying speeds. Correlation is a statistical figure calculated from the last few dozen observations, so its value drifts along with whatever is driving the market. In calm periods, when the main theme is one central bank's policy, dollar pairs can stay together for weeks. But when a local theme takes over — an election, a rate decision, one country's inflation data — a pair that was strongly correlated with others starts marching to its own beat. The sharpest changes come with crises. In a market panic, capital flees to currencies seen as safe havens, such as the dollar, franc or yen, and suddenly everything risky falls together regardless of historical relationships. This phenomenon is sometimes called "correlations going to one in a crisis", and precisely when diversification is needed most, it stops working. That is why a sensible trader refreshes their correlation table at least once a week, and checks it immediately after major macroeconomic events. For the calculation it is worth using reliable data and reading more on how markets move together in this guide to intermarket analysis.