Twin Ignition Gold
- Experts
-
Kazi Sakib Hasan
Algo trader and MQL5 developer focused on gold (XAUUSD). I build rule-based Expert Advisors with fixed lots and a stop-loss on every trade, and I publish the full backtest, including the weak years.
My product: - Version: 1.22
- Activations: 10
Twin Ignition Gold: two tested gold engines in one account, for a steadier equity curve.
Many gold EAs rely on a single idea. Twin Ignition Gold runs two independent range-breakout engines on XAUUSD in a single EA. Together they made 9.2× net profit per dollar of drawdown from 2020 to 2026, after real swap and commission, and the results held up on a second broker's data. The weaker years before 2020 are shown in full below.
Fixed lot. A stop-loss on every position. No martingale lot multiplication.
IMPORTANT: The "Minimum basket target" input is $2 per 0.01 lot. If you raise the lot size, raise this minimum in the same proportion.
BACKTEST AT A GLANCE
XAUUSD · 100% real ticks · Jan 2020 – Oct 2026 · real swap and commission charged · 0.01 lot per engine · no compounding
- Net profit: +$9,724, after $2,358 of swap and commission (+$12,083 before costs)
- Max drawdown: $1,055 (closed-trade balance)
- Net profit ÷ max drawdown: 9.2
- Profit factor: 1.19 · Trades: 8,086
- Positive months: 53 of 82 · Largest monthly gain +$980 · Largest monthly loss −$551
- Every calendar year from 2020 to 2026 was profitable:
2020 +$1,410 · 2021 +$857 · 2022 +$908 · 2023 +$324 · 2024 +$932 · 2025 +$1,514 · 2026 (Jan–Oct) +$3,779
BEFORE 2020: THE FULL STORY
The same EA and settings were also tested from January 2018. That period was hard, and we show it rather than hide it:
- 2018 (from 25 Jan): +$207 · 2019: −$1,327
- From March 2019 to April 2020 the account went 420 days without a new high
- Worst drawdown of the whole test: $1,477 equity / $1,609 balance
- Full run, Jan 2018 – Oct 2026: +$8,604 after costs, profit factor 1.14, 10,329 trades
The default settings were chosen on 2020–2026 data, so 2018–2019 is the closest thing to an unseen test, and it was the weakest part. The two MT5 Strategy Tester screenshots on this page show this full run.
TESTED ON TWO BROKERS
The same settings on a second broker's real-tick data, Jan 2020 – Oct 2026 (that broker's tester charged no swap or commission, so compare the before-cost numbers):
- Before costs: Broker A +$12,083 · Broker B +$10,294
- Longest stretch without a new high: 245 days on Broker A, 244 days on Broker B
- The worst stretch fell in the same place on both (2021–2022)
- Monthly results correlation between the two: 0.81
Different spreads and tick data give different numbers, but the shape of the results stayed the same.
HOW IT TRADES
- Mark the box. Every H4 candle becomes a range box. The quiet hours are skipped.
- Bracket it. A buy stop goes above the box and a sell stop below it. Every order carries a stop-loss at the opposite edge of the box.
- Recover. If the first breakout fails, the opposite stop takes over (ping-pong recovery). The lot stays the same every time, with a hard cap of 6 positions per engine.
- Bank the basket. When the basket reaches its ATR-based profit target, everything closes and that engine stands aside for the rest of the day.
ENGINE A — CLASSIC
Holds each position on its own range stop and recovers through ping-pong entries.
ENGINE B — RANGE RESET
Takes at most 2 entries per box. If both sides of the current box fail, it closes flat, books the loss and recovers it gradually: each following basket adds 1/4 of the outstanding loss to its normal target.
Why two engines? They trade the same boxes with different exit rules, so their weak periods only partly overlap. Tested alone on the same broker, period and costs, Engine A made 6.3× net profit per dollar of drawdown and Engine B 5.7×. Together: 9.2×, about 47% better than Engine A alone, the stronger of the two.
KEY FEATURES
- Two engines in one EA, each with its own magic number, on/off switch and rule set
- Optional time offset per engine, so an engine can run on shifted H4 boxes
- ATR-adaptive basket target (ATR(H4,120) × multiplier) with a dollar floor. It adapts to gold's changing volatility.
- Fixed lot per position. No martingale, no lot multiplication.
- Stop-loss on every position, at the opposite edge of its range
- Position cap per engine (default 6)
- Daily lockout after a winning basket, so it does not over-trade
- End-of-day sweep: pending orders are cleared before the daily rollover, when spreads are widest
- Skip-hours filter for low-quality range boxes
- Restart-safe: saves its state and resumes after a terminal or VPS restart
- Checks margin, volume limits and broker stop levels before every order
- Side-by-side dashboard for both engines plus combined P&L
- Fast Strategy Tester runs: the tester skips work that only matters on a live chart
- No DLLs, no external servers, no web requests
CHOOSE YOUR RISK
Sized from the worst drawdown in the FULL history, 2018–2026 ($1,477 equity / $1,609 balance at 0.01 lot per engine).
- LOW: $8,000 per 0.01 lot. Historical max drawdown about 18–20%. Median full year about +11%.
- MEDIUM: $5,000 per 0.01 lot. Historical max drawdown about 30–32%. Median full year about +18%.
- HIGH: $4,000 per 0.01 lot. Historical max drawdown about 37–40%. Median full year about +23%.
Median full year = 2019 to 2025, after costs (+$908 at 0.01 lot), so it includes the losing year. Scale linearly. For example, 0.02 lot per engine needs twice the balance.
REQUIREMENTS
- Symbol: XAUUSD (gold). Attach to one chart; the EA builds its H4 boxes internally.
- Account: HEDGING (the EA can hold buy and sell positions at the same time)
- Broker server time GMT+2 / GMT+3 (New York close), as used by most MT5 brokers. The skip hours and sweep times assume this.
- Low-spread ECN/raw account and a VPS recommended
- Leverage 1:100 or higher
MAIN SETTINGS
- Lot size per engine (default 0.01)
- Engine A / Engine B: on/off, rules, box offset hours, skipped box hours, magic number (defaults 770101 / 770102)
- Basket target: ATR multiplier (0.25), ATR period (120), minimum target ($2 per 0.01 lot, raise it with your lot size)
- Max positions per engine (6)
- Lockout after a winning basket (on)
- Engine B: max entries per range (2), range reset (on), loss recovery share (0.25)
- Market close time and sweep minutes
- Dashboard on/off
Default settings are the tested settings. Set your lot size, and raise the minimum basket target in the same proportion.
WHAT TO EXPECT (PLEASE READ)
- Losing years happen. 2019 lost −$1,327 at 0.01 lot per engine, and the account went 420 days without a new high (March 2019 – April 2020). Size your account so that you can sit through a period like that.
- Drawdowns happen. The worst historical drawdown was $1,477 (equity) or $1,609 (balance) at 0.01 lot per engine. Since 2020 the longest stretch without a new high was 245 days (Jul 2021 – Mar 2022).
- Returns are uneven. Gold's strong trend in 2025–2026 produced a large share of the profit. Quieter years such as 2023 earned much less.
- Costs are included. The main test charged real swap and commission ($2,358 from 2020 to 2026). Your broker's swap and commission may be higher or lower; check its XAUUSD rates.
- Up to 12 positions can be open at once (6 per engine), each at your fixed lot.
- Backtest results are not a guarantee of future performance. Trading gold with leverage carries a high level of risk and may not be suitable for all investors. Only trade money you can afford to lose.
FAQ
Is this a martingale or a grid?
No. Every position uses the same fixed lot, and every position has its own stop-loss. Recovery comes from new breakouts, not from bigger lots.
Can I run only one engine?
Yes. Each engine has its own on/off switch. In our tests, running both gave a better risk-adjusted result than either engine alone.
Can I run it on other symbols?
It was designed and tested on XAUUSD only. Our own tests on other symbols (for example GBPJPY) did not show the same results, so we do not recommend it.
Is there a free demo?
Yes. Download the free demo from this page and run it in the MT5 Strategy Tester on XAUUSD with "Every tick based on real ticks". You can test any period, including 2018–2019.
Does it need a VPS?
Strongly recommended. The EA is restart-safe, but it manages pending orders and baskets around the clock.
Will live results match the backtest?
Spread, swap, commission, slippage and broker server time all affect results. Run a demo or a small live account first, and compare it with a backtest on your own broker.
