Correlation
Correlation measures how consistently two instruments' prices move together, as a coefficient from +1, a perfect positive linear relationship, to −1, a perfect negative one.
A value of 0 means no consistent linear relationship. On price levels, +1 means the two move in step and −1 that they move in exact opposite directions. TradingView's Correlation Coefficient indicator computes the standard (Pearson) coefficient: the covariance of the two series divided by the square root of the product of their variances. It measures how consistently two prices move together, not how far, and it changes over time.
This page covers the mechanic; it is not trading advice.
What the coefficient measures, and what it does not
Positions on two different pairs are not necessarily two separate risks. EUR/USD and GBP/USD both quote in US dollars, the second currency in each Currency pair. When the dollar strengthens against both the euro and the pound, each quote falls, so a long position in each pair is partly the same position: both are short the dollar.
TradingView's Correlation Coefficient indicator compares the chart's instrument with a second symbol, using closing prices, a default length of 20 and a Source input that defaults to Close. A reading near +1 means the two closes sat above and below their window averages together, near −1 that one sat high while the other sat low, and near 0 that no linear pattern links them. On closes, a shared trend alone can push a reading towards +1, even when the two pairs move in opposite directions bar by bar.
A coefficient of 0.80 does not say that one pair moves 80% as far as the other. Multiplying one series by 2 leaves the coefficient unchanged, so a pair that always moves twice as far, in step with another, still scores +1; how far a price moves is a question of Volatility. Zero is narrower too: a series fully determined by another in a curved, non-linear way can still score 0.
Two correlated stops on EUR/USD and GBP/USD, worked
Sizing each trade at 1% sets each planned loss at the stop at 1%, not the account's total: two trades that both fill at their stops lose 2% between them, whether or not they share a currency. On a $10,000 account, each position below is sized with the Position size formula for a $100 loss at its Stop-loss; the stop distances are illustrative.
Two positions, one shared dollar leg
Combined loss if both stops trigger = Loss at stop A + Loss at stop B- Account and risk per position
- $10,000 at 1% = $100
- A: long EUR/USD 0.40 lots at 1.1253, stop 1.1228 (25 pips)
- 0.40 × 25 × $10 = $100
- B: long GBP/USD 0.25 lots at 1.3335, stop 1.3295 (40 pips)
- 0.25 × 40 × $10 = $100
Combined loss at both stops
$200 (2% of the account)
Each trade was sized for 1%; together they put 2% at risk. That maximum holds at any correlation, before trading costs and Slippage, which can fill a stop beyond its price. Correlation changes how often both stops trigger together and how widely the combined result swings. Both positions are short US dollars: buying 40,000 euros at 1.1253 sells $45,012.00, and buying 25,000 pounds at 1.3335 sells $33,337.50, $78,349.50 in total.
Assume each stop sits the same multiple of its own pair's typical move, so the two positions swing by the same cash amount. The combined swing, one standard deviation of the joint profit or loss, is then one position's swing × √(2 + 2r), where r is the correlation of the two pairs' bar-to-bar price changes, not of their closing prices; the comparison of price and change correlation below shows how far the two can differ.
Splitting one idea across two pairs looks like spreading the risk. At an illustrative r of 0.80 the two longs swing 1.90 times as widely as one position, close to the 2.00 of a single trade sized at 2%; unrelated pairs at r = 0 would swing 1.41 times as widely. With high positive correlation, the two trades behave much like one trade of twice the size.
Offsetting positions: what a 0.80 correlation leaves open
Long EUR/USD against an equal-swing short GBP/USD turns the multiplier into √(2 − 2r), with r again the correlation of price changes. At r = 0.80 that is √0.40 = 0.6325, so the pair still swings 63.25% as much as one unhedged position. The best hedge ratio, which equals r when the two swings match, leaves √(1 − 0.80²) = √0.36 = 0.60, or 60%. For a 0.80 correlation between the two prices' changes, even the best hedge ratio cuts the standard deviation of the combined result by only 40%, not 80%, and the second position pays its own Spread.
Price correlation versus change correlation
A coefficient on closing prices, TradingView's default, can differ sharply from one on bar-to-bar changes. The five closes below, starting from the example prices, are constructed, not market data.
Coefficient on the five closes
r = Covariance ÷ √(Variance of EUR/USD × Variance of GBP/USD)- Variance, EUR/USD (pips from first close: 0, 20, 30, 50, 60)
- 1,480 − 32² = 456
- Variance, GBP/USD (pips from first close: 0, 10, 30, 40, 60)
- 1,240 − 28² = 456
- Covariance
- 1,340 − 32 × 28 = 444
r on prices
444 ÷ 456 = 0.97
Variance is the average of the squares minus the square of the average; covariance is the average of the products minus the product of the averages. On bar-to-bar changes the story reverses: EUR/USD rose 20, 10, 20 and 10 pips while GBP/USD rose 10, 20, 10 and 20, and the coefficient on those changes is −1.00. The shared climb alone lifted the price reading to 0.97.
Related terms
Related articles
Common questions
What does a correlation of 0.80 mean?
A coefficient of 0.80 means two series were strongly but not perfectly linearly related over the measured window, on a scale from +1 (a perfect positive linear relationship) to −1 (a perfect negative one). On closing prices, a shared trend alone can produce a reading that high even when the bar-to-bar moves differ. It does not mean one pair moves 80% as far as the other, because multiplying either series by a positive number leaves the coefficient unchanged. Nor does it mean the pairs move the same way on 80% of bars: in a constructed five-bar example where EUR/USD and GBP/USD rose on every bar, the coefficient on those changes was −1.00.
Does trading EUR/USD and GBP/USD together spread the risk?
Not for the maximum loss, and for the swing only as far as the two pairs' correlation sits below +1. Long EUR/USD at 1.1253 and long GBP/USD at 1.3335 are both short US dollars. On a $10,000 account, 0.40 lots with a 25-pip stop and 0.25 lots with a 40-pip stop each carry a $100 loss at the stop, so the two stops together put $200, or 2%, at risk at any correlation, before slippage. At an illustrative 0.80 correlation between the two pairs' price changes and equal cash swing per position, the pair swings 1.90 times as much as one position, close to the 2.00 times of a single trade sized at 2%, against 1.41 times if unrelated.
Can a correlated pair be used as a hedge?
Only partly. Holding long EUR/USD at 1.1253 against an equal-swing short GBP/USD at 1.3335, with an illustrative 0.80 correlation between the two pairs' price changes, leaves a combined swing of √(2 − 2 × 0.80) = 0.6325, or 63.25% of the unhedged position. Even the best hedge ratio leaves √(1 − 0.80²) = 0.60, or 60%. The hedge also pays a second spread, and the correlation itself changes over time.