Zephyros trades gold breakouts with pending stop orders. It waits at a price level, and if gold breaks through,
the order fills with its stop loss and target already attached. No grid, no martingale, no averaging down. One
fixed lot, every time.

It comes with two presets, and I run both: one chart with each, on the same account. The headline number is that
pair together:

$2,000 became $6,702 in a real-tick backtest from 1 July 2021 to 16 October 2026, 0.01 lot per preset, with
commission and swap taken out. That's +235% over a little more than five years, and every calendar year in the
test finished positive.

Zephyros balance curve, real tick backtest 2021-2026

The backtest, in numbers

Zephyros' own C# engine (the same code the cTrader build runs) on Exness real ticks, XAUUSD, 2021-07-01 to
2026-10-16, $2,000 start, 0.01 lot per preset, commission $4 per lot, swap at Exness rates.

Both presets V3A-Dual alone HiWR-V3 alone
Net profit +$4,702 +$3,769 +$933
Profit factor 2.60 2.90 1.97
Trades 1,445 957 488
Win rate 82.4% 81.6% 83.8%
Max drawdown (balance) $204 (6.0%) $204 (6.6%) $68 (2.8%)

By year, both presets together: 2021 (from July) +$64, 2022 +$358, 2023 +$198, 2024 +$401, 2025 +$1,313,
2026 to mid-October +$2,369.

Zephyros result by month, 2021-2026

Why two presets

They're the same engine with two different personalities.

V3A-Dual is the earner. Wide target, and it'll keep two orders waiting on each side. Most of its trades end
close to breakeven, and then every so often one runs all the way and pays for a month. 45 of its 957 trades hit
the full target and those 45 made $5,357.

HiWR-V3 is the steady one. Tight target, one order per side, a smoother line with a drawdown of $68 across
five years. On its own it earns a lot less.

Together the profit goes up by a quarter while the worst drawdown stays where V3A-Dual's already was ($204). That's
why I'd run both. If you only want one, V3A-Dual makes the money and HiWR-V3 gives the quieter ride.

The two presets, cumulative profit

What it does, and what it doesn't

  • every order goes out with its stop loss and take profit attached, sized as a share of the gold price, so the
    distances grow with gold instead of staying stuck in dollars from 2021
  • once a trade has moved far enough in its favour, the stop jumps past the entry price. Once. No fiddly trailing
  • orders that sit untouched for too long get cancelled instead of waiting forever
  • never more than 10 open trades per preset, fixed lot, never increased after a loss

The levels it picks and the rules behind them stay private. That's the part that took the time.

Same EA on MT5, MT4 and cTrader

Zephyros was built on MT5, then ported. I checked each port by running the same months and matching trades one by
one (same direction, entry within two minutes) against the MT5 Strategy Tester on Exness real ticks, V3A-Dual
preset:

  • cTrader engine vs MT5, Nov 2025 - Oct 2026: 201 vs 199 trades, 98% matched, $2,187 vs $2,195
  • MT4 vs MT5, Jan 2025 - Oct 2026: 384 vs 388 trades, 95% matched

They're not identical. MT4 builds its ticks from one-minute bars with a fixed spread, and every broker's gold feed
is a little different. Same rules, same settings.

Downloads

One file per platform, and it's the same file whether you're testing or trading.

Backtest it for free. Anyone can download it, no account needed. It runs in the Strategy Tester (MT4, MT5) or
in Backtesting (cTrader) without a licence, with exactly the presets that trade live. Test it on your own broker's
gold before you decide anything. Give it a year or more of data: Zephyros trades about five times a week across
both presets, so a two-month test tells you very little.

Trade it with the Pro plan. When the backtest convinces you, subscribe to Pro and type your MoonTheTRAIN
username into the EA. It checks the licence online, no second download. On MT4/MT5 there's one extra step, once per
terminal: Tools > Options > Expert Advisors, tick "Allow WebRequest for listed URL" and add
https://moonthetrain.com. No DLL needed, so it also runs on the MQL5 VPS.

Try it on demo first. On a demo account Zephyros runs with any MoonTheTRAIN plan, including the free 3-day trial, so you can watch it trade with your broker's real prices before you pay for Pro.

New versions. When I release an update, Zephyros tells you: a line on the chart, a popup, and on MT4/MT5 a push
to your phone if you've set up notifications. Download the new file here and swap it in.

Set-up

  • Symbol: XAUUSD, or whatever your broker calls gold (XAUUSDm, GOLD...). It refuses other symbols.
  • Charts: two, any timeframe. Preset V3A-Dual on one, HiWR-V3 on the other. Leave the magic base (or label
    prefix on cTrader) the same on both: the EA adds the preset to it, so the two charts never touch each other's
    orders.
  • Account: hedging recommended. The two presets can hold opposite trades.
  • Capital: $2,000 per 0.01 lot per preset is what the test above used. Not less.
  • VPS: recommended. Pending orders are checked around the clock.

Inputs you can change

Input Default What it does
Preset V3A-Dual V3A-Dual or HiWR-V3. Run one of each on two charts
Lot per trade 0.01 Fixed size for every trade
Magic base (MT4/MT5) / Label prefix (cTrader) 2026110 / Zephyros Keeps Zephyros apart from other EAs; the preset number is added to it
Broker server time zone (MT4/MT5) Auto Leave on Auto live. In the tester pick Fixed 0 for Exness or "GMT+2/+3" for most other brokers
Server GMT offset (MT4/MT5) 0 Used with Fixed mode in the tester

Everything else (levels, targets, stops, breakeven, expiry) is fixed at the values in the test. Those are the
settings the numbers above belong to.

Questions people ask

Can I run just one preset? You can. V3A-Dual alone made $3,769 in the test with the same $204 drawdown. I run
both because HiWR-V3 adds profit without adding to the worst drawdown.

Does it trade on news? It doesn't read a calendar. Orders are already sitting at their levels, so a news spike
that breaks one will fill it. The stop loss is attached from the start.

Which timeframe? Any. It builds the candles it needs on its own.

Past results don't guarantee future results. Test on a demo account first.