Four open trades do not necessarily represent four separate risks. A trader buying EUR/USD, GBP/USD and AUD/USD while selling USD/CHF has several positions, but every one includes exposure to a weaker US dollar. One adverse move could hurt the entire group.
Correlation measures how returns move together. Hidden concentration is the broader problem: positions that look different can depend on the same currency, sector, economic outcome or trading condition. In speculative trading, the useful question is not simply how many trades are open. It is how much money could be lost if one shared assumption proves wrong.
What Correlation Measures—and What It Misses
The commonly used Pearson correlation coefficient measures the strength and direction of a linear relationship. It ranges from −1 to +1: positive values indicate returns tending to move together, while negative values indicate returns tending to move in opposite directions. A value near zero indicates little linear relationship over the measured sample, not proof that the positions are independent. These distinctions follow the NIST definition of the correlation coefficient.
A correlation of +0.8 does not mean two markets move in the same direction on 80% of days. Nor does it mean their moves are equally large. One instrument might fluctuate much more sharply than the other while maintaining a strong positive relationship.
For trading decisions, distinguish between the correlation of market returns and the correlation of your positions’ profits and losses. Two positively correlated markets may partly offset each other if you buy one and sell the other. Conversely, buying one market and selling a negatively correlated market can reinforce the same risk.
Position direction, size and volatility all matter. A correlation grid without those details describes relationships between markets, not the amount your account could lose.
Where Hidden Concentration Appears
Several currency pairs, one dollar position
Currency trades contain two exposures. Buying GBP/USD means taking a long sterling position against the dollar; selling USD/CHF means taking a short dollar position against the Swiss franc. Different pair names do not remove the shared dollar exposure.
Consider this hypothetical £20,000 account. Each position has a planned £100 loss from entry to its stop, excluding costs and execution differences.
| Position | Currency exposure | Planned loss at stop |
|---|---|---|
| Buy EUR/USD | Long euro, short US dollar | £100 |
| Buy GBP/USD | Long sterling, short US dollar | £100 |
| Buy AUD/USD | Long Australian dollar, short US dollar | £100 |
| Sell USD/CHF | Short US dollar, long Swiss franc | £100 |
Each trade risks a planned 0.5% of account equity. Together, they carry £400, or 2%, of planned loss if all four reach their stops. They need not move identically or hit their stops together. Nevertheless, a broad dollar rally is a plausible common threat.
The £400 figure is not a statistical loss forecast. It is a straightforward exposure check: four trades that may lose from the same event have consumed four portions of the risk budget.
Different instruments, overlapping holdings
Concentration also appears when a trader holds individual technology shares alongside a technology fund and a broad equity index containing the same companies. The wrappers differ; some underlying exposures repeat. FINRA’s guidance on concentration risk identifies both correlated holdings and overlapping fund investments as reasons a portfolio may be less diversified than it appears.
For a trading account, look through the product name to the underlying exposure. A share position, an index contract and a sector fund might all suffer under a scenario involving falling technology valuations. Treat that scenario as a potential risk group rather than assuming three product types provide three independent outcomes.
Different strategies, the same failure condition
Strategy labels can hide overlap too. Suppose three systems use different entry indicators, but each buys equity index breakouts and exits after a sharp reversal. Their rules differ, yet the same choppy session could hurt all three.
Group strategies by what makes them lose, not just what makes them enter. Ask whether they depend on persistent trends, stable price relationships, calm volatility or reliable execution. This turns an abstract correlation question into a practical one: what would have to happen for several systems to fail together?
Why Trade Size Matters as Much as Correlation
Correlation alone cannot tell you which position dominates account risk. A small, volatile trade may contribute more to daily profit and loss than a much larger position in a calmer market. Equal cash allocations are not necessarily equal risk allocations.
Consider a simplified mathematical example. Suppose two positions each have a daily profit-and-loss standard deviation of £100. If their correlation is zero, the standard deviation of their combined result is approximately £141. At a correlation of +0.8, it rises to approximately £190. At +1, it reaches £200.
Those figures follow the two-position relationship:
Combined variance = first variance + second variance + 2 × correlation × first standard deviation × second standard deviation.
This calculation describes variability under the stated assumptions. It does not establish a maximum loss, predict tomorrow’s outcome or account for changing market conditions. The £100 figures are standard deviations, not stop distances or amounts staked.
The practical implication is simple: adding a strongly correlated position can add almost as much variability as enlarging the original trade. Use position sizing and a trading risk budget to decide whether the combined exposure is acceptable before treating the second entry as a separate opportunity.
How to Measure Correlation Across Your Trades
Use returns over matching intervals
For a basic market comparison, calculate percentage returns over the same intervals rather than correlating raw price levels. Two price series can both trend upwards without their period-by-period returns providing useful diversification.
Align timestamps, trading sessions and missing observations. Comparing one market’s closing price with another market’s price several hours later can distort the relationship you are trying to measure. Where holdings involve different currencies, also inspect results converted into the account currency.
The data interval should reflect the decision. Daily returns can provide context for a position held for weeks, but they should not be the only evidence used to assess trades that overlap for twenty minutes around an announcement.
Compare several observation windows
As a working check, compare a shorter window, such as 20 trading days, with a longer window, such as 120 days. These are illustrative choices, not prescribed settings. A shorter sample reacts faster but contains fewer observations; a longer sample includes more history but may blend different market conditions.
Rather than reducing everything to one coefficient, inspect rolling readings and the underlying return chart. Ask whether a relationship is persistent, recently changing or dominated by a handful of unusually large moves.
Measure strategy results as well as market returns
For multiple strategies, compare profit and loss over matching time intervals, including open-position changes where possible. Looking only at completed trades can miss simultaneous floating losses because positions may close at different times.
Keep scaling consistent. Separate changes in strategy behaviour from changes caused by larger position sizes. Then inspect the combined account result after realistic costs. The workflow in building and testing a forex trading strategy provides a useful framework for testing trading rules without treating historical results as promises.
Why Historical Diversification Can Disappear
Correlation is an estimate from a sample, not a permanent property of two markets. An economic shock can change which forces dominate prices and weaken a relationship that previously reduced portfolio risk.
A documented example is the change in US equity and government bond returns around mid-2021. Their correlation shifted from negative to positive as the inflation environment changed, weakening the protection bonds had provided against equity losses. The BIS research on equity and bond return correlation, published in December 2023, examines that shift. It is historical evidence of changing relationships, not a claim about today’s correlation.
This does not mean every correlation becomes +1 during a crisis. That slogan replaces one unsafe assumption with another. Instead, test the particular way your diversification could fail: an expected hedge might fall alongside the position it was intended to protect, or provide too little profit to offset the loss.
Market functioning deserves a separate check. Stress can involve thinner liquidity and disrupted pricing relationships, not just larger price changes. The BIS analysis of market conditions under pressure examines volatility, illiquidity and departures from normal pricing relationships across major markets.
For your own stress test, specify an adverse event, estimate each position’s response and total the account impact. For the currency example, that might mean a broad dollar rally with different moves in each pair. The purpose is not to forecast the event precisely, but to expose a loss the ordinary correlation estimate may conceal.
Set Limits for Shared Risks, Not Just Individual Trades
A per-trade limit controls one entry. A group limit controls several entries exposed to the same adverse event. Both are useful, but neither replaces an account-wide limit.
Suppose the hypothetical £20,000 account uses a £200 planned-loss ceiling for one currency theme. Four dollar-sensitive setups cannot each receive £100 of risk without exceeding that ceiling. The trader could select fewer setups, reduce their sizes or decline further exposure. The £200 ceiling is an illustration, not a recommended percentage.
Do not shrink the recorded risk simply because a recent correlation reading looks reassuring. First check the combined planned loss without assuming offsets. Then assess any diversification benefit separately, including what happens if it disappears.
Planned stop losses also need an execution allowance. For ordinary stock stop orders, the trigger price is not a guaranteed execution price; a stop-limit order introduces the possibility of no execution instead. The SEC investor bulletin on stop and stop-limit orders sets out those distinctions. Check the actual order terms for each product rather than treating every stop as a fixed loss cap.
Where several trades share an announcement risk, assess them together before the release. A separate entry signal does not create a separate economic event. The guide to news trading and execution risk covers the interaction between surprises, price moves and execution.
Hedging also requires care. Buying one bank share and shorting another may reduce broad sector exposure, but it leaves a relative-performance trade. Different position sensitivities can produce an imperfect offset. Do not subtract two equal cash amounts and assume the remaining risk is zero.
A Practical Pre-Trade Concentration Check
Before adding a position, write down its direction, size, underlying market, planned loss and most plausible adverse event. Compare it with what is already open, including positions held through other accounts.
- Identify the shared driver. Could the same currency move, sector decline or announcement hurt existing positions?
- Total the exposure. Add planned losses for the affected group, then inspect a harsher execution scenario.
- Challenge the offset. If another position is supposed to protect the account, test what happens when it provides less protection than expected.
- Make the size decision. Accept, reduce, replace or reject the proposed trade before placing it.
Repeat the check when positions grow, new trades are added or the reason for holding them changes. Record which groups actually lost together, then compare those outcomes with the assumptions used before entry.
The objective is not to eliminate every correlated position. It is to make concentration deliberate, measurable and affordable. Several trades may express one sensible view—but the account should be sized for one shared failure, not protected by the illusion of several independent bets.