A spread between pairs: trading the residual

[Fairman]

A spread between pairs: trading the residual

EURUSD and GBPUSD move together, but not in lockstep. The difference between them, adjusted for their usual relationship, wanders around a mean. When it stretches, a reversion trade is long one pair and short the other.

The appeal is neutrality. The dollar's influence cancels out, leaving a bet on the relative performance of sterling and the euro.

For scalping, the issues are cost and execution. Two legs mean two spreads on entry and two on exit. The fills aren't simultaneous, so there's leg risk: one fills and the other moves away.

The relationship also shifts. News specific to one currency breaks the pattern. A UK announcement can send the residual far from its mean and keep it there.

A rolling regression estimates the relationship. A fixed ratio doesn't hold for long. The hedge ratio must adapt, which adds estimation error.

The returns are small, a few pips per round trip, and costs are substantial. Professionals do this with much lower costs and faster execution.

For retail scalpers, the strategy mostly exists on paper. Where it can work is in slower versions, with targets of 10 to 20 pips and holding times of hours.

After both spreads, what residual strategies, if any, have worked?

[Amanhasnoname]

I ran a EURUSD versus GBPUSD residual model on five-minute bars for most of a year, and the honest answer is that the fast version never survived costs for me. Paying four half-spreads plus a bit of slippage on the second leg ate nearly everything the reversion produced, and the leg risk was worse than I expected because cable would often move while my euro fill was still confirming.

What did work, modestly, was the slower approach you describe. I used a 60-day rolling regression for the hedge ratio, only entered when the residual was beyond two and a half standard deviations, and aimed for the residual to come back halfway. Holding times were three to eight hours, and the target was around 15 pips net across both legs.

The thing I'd add is a hard rule for regime breaks. Whenever there was a Bank of England surprise the relationship shifted for weeks, so I paused the model for ten trading days after any BoE decision that moved cable more than 80 pips.

[Shadow Trader]

Some scalpers trade a single pair for years and know its personality: when it spikes, how it reacts to news, what its spread does at 3pm. Others rotate through six or eight pairs looking for the best setup.

Specialists have an advantage in pattern recognition and in understanding the pair's quirks. Switching pairs means constantly re-learning, and the quirks of a thin pair are very different from those of EURUSD.

Rotators have more opportunities and can dodge pairs in unfavorable conditions. They also face higher cognitive load, more monitoring, and often worse costs since many of those pairs have wider spreads.

Correlation complicates the diversification argument. Trading EURUSD, GBPUSD, and AUDUSD at the same time during a dollar move is a single bet repeated three times.

The data from a trade log, results by pair, shows which approach fits. Many traders discover that a majority of profit comes from one or two pairs, and the rest dilute it.

A middle route: focus on one pair as the primary, with a second as a backup when conditions turn poor.

Which pair produces most of the profit in the trade log, and what is the share of total trades?