Skip to content
Retired — tested by us Our validations

The FX Carry Trade (G10)

Borrow low-yield currencies, hold high-yield ones, collect the rate differential.

Failed honestly on 15 years of real data — the rate-ranked signal lost to a random-sign placebo.

Why it fails
Over 2011–2026 the interest differential earned a thin trickle while the FX leg bled ~4× as much; the real strategy underperformed random.
When / how it stopped
The G10 carry premium compressed after the 2008 ZIRP era. Our hardened v2 across 2011–2026 (incl. the 2015 CHF de-peg and COVID) posts PF 0.844, Sharpe −0.29, −44.5% drawdown.

The carry trade is the most famous currency strategy there is: hold the high-interest-rate currencies, fund them with the low-rate ones, and pocket the difference. It has a real academic pedigree — but that pedigree is mostly pre-2008 and mostly emerging-market baskets.

We ran a hardened, pre-registered version on 15 years of real G10 data (2011–2026) through all 11 gates. It failed decisively:

  • Profit factor 0.844, Sharpe −0.29, max drawdown −44.5% — final equity −32%.
  • The decisive result was the placebo: a third of random currency-sign assignments beat the real rate-ranked strategy, which sat below the placebo 95th percentile. Ranking currencies by interest rate was worse than random over this window.
  • Decomposition: carry earned +$848, the FX leg lost −$2,261. The premium was real but tiny; the currency moves swamped it.

An earlier 3.5-year version (2023–2026) looked deployable at PF 1.17 — which is precisely why a single favorable window is not evidence. Extend the sample and the edge evaporates.

This is not a contradiction of the carry literature; it is consistent with the well-documented “lost decade” for G10 carry. The premium compressed after the global ZIRP era while the canonical carry crashes kept happening. For a $5k retail account after costs, it does not clear the bar.

The full pre-registered report, gate scorecard and charts are in our validation write-up.

→ Read our full validation report: /strategy/carry-hardened-v2

Sources

Frequently asked

Does the FX carry trade still work in 2026?

Not on G10 majors for a retail account. Tested through 11 pre-registered gates over 2011–2026, hardened cross-sectional carry returned a profit factor of 0.844 (below 1.0), a Sharpe of −0.29, and a −44.5% max drawdown — and the rate-ranked signal was beaten by a third of random sign-permutations. The premium that exists academically lives mostly in pre-2008 data and in broad emerging-market baskets, not post-2008 G10.

Why did the carry trade stop working?

The near-universal zero-interest-rate era after 2008 compressed the differentials carry feeds on, while the currency leg kept taking carry-crash losses (the 2015 SNB franc de-peg, the 2014–15 dollar super-cycle). When the premium you harvest shrinks and the tail losses do not, the math turns negative.

Not investment advice — your mileage may vary, but the burden of proof is on the person claiming an edge. This entry describes general research and published evidence (or its absence), not a recommendation. See the full disclaimer.