Is hedging correlated pairs a free lunch?
Posted: Tue Oct 27, 2026 7:01 am
The pitch: buy EURUSD and sell GBPUSD, since they're highly correlated. If the dollar moves, one gains and the other loses, so risk is neutralized. Profit from the difference. What could go wrong?
Plenty. The correlation isn't perfect, and it changes. When the pair diverges, both legs can lose at once. Each leg pays its own spread and commission, doubling the cost. And a hedge doesn't produce profit, it only reduces exposure to the common factor, leaving a spread trade that needs its own edge.
Most so-called hedges are really two directional bets that happen to partly offset. The net exposure might be a long euro, short pound position, which is a view on EURGBP. That could simply be traded directly at a single spread.
There's nothing wrong with relative value trades, but they should be understood as such, with their own risk and costs.
If a hedge is meant to carry a view on the spread between two pairs, it's worth calculating the spread series itself and testing whether it mean-reverts or trends. Without that, the position is just a more expensive way of holding two opinions.
Has a correlated-pair hedge ever worked as intended, or did it end up as two losing legs?
Plenty. The correlation isn't perfect, and it changes. When the pair diverges, both legs can lose at once. Each leg pays its own spread and commission, doubling the cost. And a hedge doesn't produce profit, it only reduces exposure to the common factor, leaving a spread trade that needs its own edge.
Most so-called hedges are really two directional bets that happen to partly offset. The net exposure might be a long euro, short pound position, which is a view on EURGBP. That could simply be traded directly at a single spread.
There's nothing wrong with relative value trades, but they should be understood as such, with their own risk and costs.
If a hedge is meant to carry a view on the spread between two pairs, it's worth calculating the spread series itself and testing whether it mean-reverts or trends. Without that, the position is just a more expensive way of holding two opinions.
Has a correlated-pair hedge ever worked as intended, or did it end up as two losing legs?