Combining Strategies: Diversification That Actually Works
Running two strategies can smooth your results, but only if they behave differently. Two strategies that win and lose together provide no diversification.
The key measure is correlation of returns. If one strategy trades trend breakouts on EUR/USD and another trades trend pullbacks on EUR/USD, they will often lose in the same choppy periods. A trend strategy and a mean-reversion strategy tend to complement each other, since one thrives when the other struggles.
Diversifying across pairs helps, but beware that many pairs share the dollar. Five dollar pairs may be one bet. Crosses and different currency groups add real variety.
Timeframes and sessions provide another dimension. A swing strategy on daily charts and an intraday strategy in the London session are unlikely to be perfectly aligned.
Allocate risk, not just capital. Decide the risk per trade for each strategy so that combined exposure stays within your total open risk cap.
The costs are complexity and attention. Each strategy needs its own tracking, review, and discipline. Beginners often benefit from mastering one approach before adding another.
Check the combined equity curve. A good combination has shallower drawdowns than either component alone.
Do not add a strategy just to feel busy.
Practical step: line up the monthly results of two approaches and see whether their good and bad months overlap.
Combining Strategies: Diversification That Actually Works
Re: Combining Strategies: Diversification That Actually Works
The five dollar pairs point is so common. Buying euro, pound and aussie against the dollar at the same time can feel like three ideas, but it's really one bet against the dollar at three times the risk.
It’s Fairman 
Re: Combining Strategies: Diversification That Actually Works
Agree with mastering one first. Adding a second setup before the first is solid mostly doubles the journal work and makes it harder to tell which one is actually losing.
It’s Fairman 
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: Combining Strategies: Diversification That Actually Works
Lining up monthly results is a good start, but a year gives only twelve pairs of numbers, and a correlation from so few points can swing wildly. Weekly R for each strategy over a year gives 52 points, still small but far steadier.
The correlation that matters most is the one during losing stretches. Two strategies can show near zero correlation across the whole year and still lose together in the worst weeks, often around big news or a sudden jump in volatility.
A quick test is to take the ten worst weeks of the main strategy and check what the second one did in those same weeks. Say it was flat or positive in seven of them, then it is doing its job, even if the headline correlation looks unremarkable.
On risk allocation, I would argue against an even split at the start. Giving the newer strategy a quarter of the risk until it has 100 live trades keeps a weak second system from dragging down a proven first one.
How do people here handle a trend setup and a mean reversion setup on the same pair firing in opposite directions at once? Netting them seems sensible, but it hides which one would have won.
The correlation that matters most is the one during losing stretches. Two strategies can show near zero correlation across the whole year and still lose together in the worst weeks, often around big news or a sudden jump in volatility.
A quick test is to take the ten worst weeks of the main strategy and check what the second one did in those same weeks. Say it was flat or positive in seven of them, then it is doing its job, even if the headline correlation looks unremarkable.
On risk allocation, I would argue against an even split at the start. Giving the newer strategy a quarter of the risk until it has 100 live trades keeps a weak second system from dragging down a proven first one.
How do people here handle a trend setup and a mean reversion setup on the same pair firing in opposite directions at once? Netting them seems sensible, but it hides which one would have won.
-
FXS Prop Desk
- Posts: 1689
- Joined: Wed Oct 07, 2026 9:52 am
Re: Combining Strategies: Diversification That Actually Works
A practical check for diversification is to compute the correlation of the two strategies' daily returns over the test period. Values close to zero or negative indicate real diversification. Also looking at whether their worst drawdowns happened simultaneously gives a more complete picture than the correlation coefficient alone.
Combining strategies on different pairs also helps, as long as the pairs themselves are not highly correlated, as EUR/USD and GBP/USD often are.
Combining strategies on different pairs also helps, as long as the pairs themselves are not highly correlated, as EUR/USD and GBP/USD often are.
FXS Official Team | The Idea Lab & Strategy
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: Combining Strategies: Diversification That Actually Works
Daily correlation has a quirk with strategies that trade only a few times a week. On days when one of them has no trade its return is zero, and a series full of zeros pulls the correlation towards zero even if the two lose together whenever both are active.
Say each strategy trades on about a third of days, then most pairs of daily figures include at least one blank. A cleaner test is to look only at days where both had a position, even if that leaves fewer points.
Holding time adds a second wrinkle. A scalping method closes within the day while a swing method carries risk across several, so lining them up by close date puts the whole swing loss on a single day it did not really happen on.
Marking each open position to the close of every day fixes that, though it takes a little spreadsheet work. Without it the swing strategy looks smoother and less connected to the scalper than it really is.
On FXS Prop Desk's EURUSD and GBPUSD example, their correlation is not constant either. It tends to drop around UK data, which is exactly when a trader running both might be counting on it.
Say each strategy trades on about a third of days, then most pairs of daily figures include at least one blank. A cleaner test is to look only at days where both had a position, even if that leaves fewer points.
Holding time adds a second wrinkle. A scalping method closes within the day while a swing method carries risk across several, so lining them up by close date puts the whole swing loss on a single day it did not really happen on.
Marking each open position to the close of every day fixes that, though it takes a little spreadsheet work. Without it the swing strategy looks smoother and less connected to the scalper than it really is.
On FXS Prop Desk's EURUSD and GBPUSD example, their correlation is not constant either. It tends to drop around UK data, which is exactly when a trader running both might be counting on it.