Stairway to Heaven: a trend-following breakout system on gold. 31% win rate.
Lurked here for years, finally have something worth throwing to the wolves.
Pure trend-following: Donchian-style breakout entry, single fixed trailing stop on the exit. No grid, no martingale, no averaging down. One position at a time, fixed risk, and it either trails into a move or gets stopped out small. I called it Stairway to Heaven before I had a good reason to, then the equity curve grew stairs, so now it's earned.
The run below is XAUUSD, ~18 months (Jan 2025 → Jun 2026), real-tick backtest at 100% quality, $10k start:
- Net +$18.9k (+188%)
- Profit factor 1.74, Sharpe 2.89, recovery factor 3.10
- Max drawdown ~21.41%
- 1,136 trades, Win rate: 31%. Average win $126, average loss $33.
Let me get ahead of the top comment: yes, it loses ~7 trades out of 10, on purpose. It bleeds a thin stream of small stops through the chop (those flat, sagging stretches on the curve are it paying rent) and then catches the occasional real trend and lets it run.
The whole edge lives in that ~3.8:1 win/loss ratio, not in the entry. Honestly the hardest part isn't the code, it's sitting through six weeks of slow bleed without touching anything. If watching your equity leak sideways makes you itch, this style will eat you alive. That discomfort is the moat.
Things I think actually matter here:
- Peak exposure is ~1% of the account (bottom subgraph). No leverage tricks, no "hold and pray it comes back." Worst case per trade is a known, fixed stop.
- It's long-biased in the stats (longs win 35%, shorts 27%), which lines up with gold's regime over this window — so some of this is the market, not me.
- That big step in March 2026 is the gold spike. A trend system's dream. But look right after: it hands a chunk back and grinds sideways for months. There is no free lunch, only a delayed one.
Caveats, because I'm not selling anything:
- One instrument. I have no illusion this generalizes for free, trend systems are famously regime and instrument dependent.
- It's a backtest. Real ticks help, but breakout fills are exactly where live slippage bites, and I haven't stress-tested that properly yet.
What I'd genuinely like input on:
- Single trailing stop vs scaling out, every partial-TP variant I test lowers my Sharpe, which surprised me. Anyone found the opposite?
- How do you personally survive the flat stretches in a live account, sizing rules, a basket of uncorrelated systems?