What Is Correlation Risk in a Trading Portfolio?

In short. Correlation risk is the risk that several open positions lose money at the same time because they respond to the same driver. Positions that each look small can add up to one large bet when they move together. It is measured with the correlation coefficient, from −1 to +1: the closer to +1, the less the positions diversify each other. Correlations also tend to rise in falling markets, exactly when diversification is needed most.

Most traders size each trade on its own: a stop, a risk of 1% of the account, a position size calculated from the distance. That logic is sound for one trade. It stops being enough when three or four trades are open at once and all of them depend on the same thing, such as the US dollar, the crypto market as a whole or one sector's earnings. Correlation risk is the name for that hidden overlap.

What correlation risk means

Two positions are correlated when their results tend to move in the same direction at the same time. Correlation risk is what happens to the account when that tendency is strong: a single move in the common driver hits every position together, and the total loss is closer to the sum of all the stops than to the loss of any one trade.

The opposite case exists too. Positions with low or negative correlation tend to offset each other, so the account swings less than the individual trades would suggest. That is the whole idea behind diversification, and it only works when the correlation is actually low, not when the instruments merely have different names.

How correlation is measured

The standard measure is the linear correlation coefficient, usually written r. The IUPAC Gold Book defines it as a measure of the degree of interrelationship between two measured quantities, calculated from how far each pair of values deviates from its average. In trading, those quantities are usually the daily or per-trade returns of two instruments or two strategies.

The ranges in the middle rows are common rules of thumb, not official thresholds. Two limitations matter for traders. The coefficient only captures linear relationships, and it depends heavily on the period it was measured over: a 30-day window and a 2-year window can give very different values for the same pair.

Why correlated positions add up

When positions are independent, their random swings partly cancel. When they are correlated, they do not. The standard portfolio formula behind this goes back to Harry Markowitz's 1952 paper "Portfolio Selection": the variability of a combined position depends not only on the size of each part but on the correlation between them.

The table below shows our own calculation for a simple case. Each open position has the same typical swing, equal to 1% of the account. The column on the right shows the "effective number of independent bets", a way to see how much diversification the trader actually has.

These figures are illustrative calculations, not data from real accounts. The pattern is the useful part: with five positions at a correlation of 0.8, the trader holds five tickets but only a little more than one independent bet, and the account swings almost as if one position five times the normal size were open.

The worst case is simpler still. If every stop is hit in the same move, the loss is the sum of the planned risks, whatever the correlation was in calm periods. With five positions at 1% each, that is 5% in one move.

Correlation is not stable

A correlation measured over a quiet period is a weak guide to how positions will behave in a sell-off. François Longin and Bruno Solnik studied monthly returns of the five largest stock markets from 1958 to 1996 in "Extreme Correlation of International Equity Markets". They found that correlation is related less to volatility as such and more to the market trend: it increases in bear markets, but not in bull markets.

The same paper warns that simple comparisons can mislead, because correlation measured only on large moves looks higher even when the underlying relationship has not changed. For traders, the practical reading is cautious: diversification that looks fine on a calm chart may shrink when several positions go wrong at once.

Common sources of correlation risk

Correlation risk rarely looks like "the same trade twice". It usually hides behind different tickers:

• A shared currency: long EUR/USD, long GBP/USD and short USD/CHF are three positions, but all three are bets against the US dollar.
• One market, many coins: long positions in several altcoins often depend on the direction of the crypto market as a whole, led by Bitcoin.
• Index and its components: a long index future plus long positions in its largest stocks doubles the exposure to the same market move.
• One sector or theme: several stocks from the same industry tend to react to the same news, earnings season or rate decision.
• Several prop accounts, one strategy: copying the same trades across accounts does not spread risk; one bad day can hit every account at once.

The exact strength of each relationship changes over time, so the list above is a set of questions to check, not a set of fixed numbers.

How to manage correlation risk in practice

Correlation risk is managed at the level of the group, not the individual trade. A workable routine looks like this:

1. Group positions by driver: before entering, ask what single move would make this trade and the open ones lose together.
2. Cap the risk of each group: set a limit for combined risk across positions that share a driver, and size each new trade within what is left of it. The broader framework is in our guide to risk management for traders.
3. Size from the group, not only the trade: standard position sizing answers "how big is this trade"; a group cap answers "how big can this trade be, given what is already open".
4. Check against account limits: on accounts with a daily loss limit, the sum of stops in one correlated group should stay below that limit with a margin.
5. Measure on your own data: calculate correlations from your journal's actual returns over a fixed window and recheck them at each review.

Correlation also affects formulas that assume independent trades. The Kelly criterion and a Sharpe ratio built from several strategies both look better on paper when the overlap between positions is ignored.

What to log to see correlation risk

Correlation risk is hard to see trade by trade and easy to see in aggregated records. Useful fields:

• Entry and exit time for each trade, to see which positions were open at the same moment.
• Planned risk in account currency, so the combined risk of overlapping positions can be added up.
• Instrument, market and main driver (for example "USD", "crypto beta", "tech sector") as a tag.
• Strategy and account, to check whether several strategies or accounts are really one exposure.
• Daily results, to spot days when several positions lost together, which is the clearest sign of correlation in practice.

This article is for educational purposes only and is not financial or investment advice. The figures are illustrative calculations, not recommendations. Trading, especially with leverage, carries a high risk of losing some or all of your capital, and correlations measured in the past do not guarantee how positions will behave in the future.

A trading journal that records risk, instrument and strategy on every trade, together with a performance dashboard that shows results by day and by account, makes it easier to see when several positions are really one bet.