Daily Trend-Following (50/200 SMA)
Go long above the 200-day moving average / on the golden cross, short below — the classic trend filter.
Only gold made money; all seven FX majors lost — the majors ranged rather than trended in 2018–2026, so a trend system had nothing to capture.
- Why it fails
- The premise — that the major-FX basket trends — was false for this decade; every parameter pair and both out-of-sample halves lose.
- When / how it stopped
- Run on real 2018–2026 data across 8 majors plus gold, the 50/200 crossover posts net PF 0.94, Sharpe −0.02, and a −42.3% drawdown.
The 50/200 SMA crossover is the most famous trend filter in retail trading: go long when the fast average crosses above the slow one (the golden cross), short when it crosses below (the death cross). It has a genuine academic pedigree in older equity data.
Run honestly on real 2018–2026 data across eight FX majors plus gold, it loses:
- Net profit factor 0.94, Sharpe −0.02, −42.3% max drawdown, win rate 36.7% across 934 trades — the account ends at $4,247 (−15%).
- The cause is structural, not cost-related: of the eight instruments, only gold (+$2,157) made money. All seven FX majors lost. Strip gold and the FX book loses ~$3,000 on its own.
The robustness sweep confirms the edge is absent, not fragile. Every crossover pair — 50/200, 100/200, 50/100, 20/100 — lands at PF 0.91–0.99, and both out-of-sample halves lose independently (train PF 0.96, validate PF 0.97). There is no parameter island of profitability to overfit to.
To separate implementation from premise, we also tested the charitable textbook variant — always in-market, hold until the cross flips, stops removed. It reaches PF 1.01 in-sample (break-even) by killing whipsaw, then collapses to PF 0.72 out-of-sample. Even that break-even is entirely gold papering over NZD and CHF losses.
The diagnosis: the majors ranged this decade, so a trend system had nothing to capture. The transferable lesson is to pivot the instrument set, not the signal.
The full report, attribution table and equity charts are in our validation write-up.
→ Read our full validation report: /strategy/d1-trend-follower
Sources
- Our validation report — D1 Trend-Following (50/200 SMA), PF 0.94
- Brock, Lakonishok & LeBaron (1992), "Simple Technical Trading Rules and the Stochastic Properties of Stock Returns", Journal of Finance
- Sullivan, Timmermann & White (1999), "Data-Snooping, Technical Trading Rule Performance, and the Bootstrap", Journal of Finance
Frequently asked
Does trend following with the 50/200 SMA work on forex majors?
Not in 2018–2026. The classic 50/200-day crossover on eight FX majors plus gold returned a net profit factor of 0.94, a Sharpe of −0.02, and a −42.3% max drawdown. Of the eight instruments only gold made money (+$2,157); all seven FX majors lost. The majors ranged rather than trended, so a trend-following system had nothing to capture and bled whipsaw plus spread.
Is the 200-day moving average profitable, or is the loss just costs?
It was not profitable here, and the loss is structural rather than cost-driven. Every parameter pair tested — 50/200, 100/200, 50/100, 20/100 — lands at PF 0.91–0.99, and both out-of-sample halves lose. Even the most charitable textbook variant (multi-month holds, no stop-churn) only reaches break-even in-sample and collapses to PF 0.72 out-of-sample, because the underlying premise — that the majors trend — was false for this decade.
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.