Ouro EA

HOW IT WORKS

The levels. Each session Ouro reads gold's recent structure (yesterday's high/low, the extremes of the last 12 hours and two days) and keeps only the places where several extremes cluster near a round number. A handful exist at any time and some days none.

Pending orders, the dragon waits. At each level it places a stop order just beyond it (Buy Stop above resistance, Sell Stop below support) and waits. If gold breaks through, the order fills at the moment of the break. If it never gets there, the order is withdrawn before the daily close and new levels are computed next session. 

No fixed Take Profit, a staged exit instead. A fixed target is too small on $200 days and never reached on $30 days. Ouro adapts to what the trade does:

- Partial at about +2 R: closes roughly a third and secures the rest, from here the trade cannot become a full loss.

- Trailing stop from about +3.5 R: follows price at a distance, so a $100 run is allowed to run.

- Add-on on a winner: a small extra piece (~30 %) sharing the same stop.

- Time limit: 24 h (Sniper) / 8 h (Runner), no capital stuck in a dead trade.

That's why OURO EA can achieve huge profits in just one trade. 

Many small protected exits, a few very large ones. That is the profit factor.vTwo engines. Sniper: fixed $5 stop, large target, arms only when daily range ≥ $60, the big trades. Runner: stop scaled to the day's range, faster trail, trades every regime. Their monthly results are almost uncorrelated (0.14), so the combined curve is smoother than either alone. One position per engine, plus the add-on.

When it does NOT trade. Daily range below ~$60 (Sniper rests) · 30 min before to 45 min after high-impact USD/EUR news (orders pulled) · before the daily close (all pending orders withdrawn).

The panel. State (gold ARMED · jade RESTING · ember BLOCKED · grey out of window), levels in play, and Execution Quality: trade by trade, how far your fill and your stop exit were from the intended price, as % of the $5 stop. Below 8 % healthy · 8–15 % watch · above 15 % your broker is eating the edge. With a $5 stop, 30 cents of slippage is 6 % of your risk — no backtest can tell you that; the panel does, on your account.

- No grid. No martingale. No averaging down. No hedging.

WHY OURO, what typical gold robots get wrong.

We bought the best-selling gold breakout EAs and tested them on real ticks. The same problems kept coming up. Ouro was built around them.

- The 95 % win rate. They hold for seconds and lock in +$0.50; the one $150 move of the month is closed at +$0.50. → Ouro's exit is built for that move. Its win rate is deliberately lower because losers are fixed and winners are allowed to grow.

- Fixed-dollar stops and targets. "$2 stop / $15 target" is one system at $2,000 gold and another at $4,000. That is why so many gold EAs sat flat 2023–early 2025 and then "woke up". → Ouro measures the daily range every day and rests when it is not there.

- Entering before the break. Orders placed inside the level get filled by every fake-out. → Ouro enters on the break. In our tests, about 3× the return per unit of drawdown.

- Five to eight positions at once. One bad hour hits all of them. We measured −20 % days at the advertised 2 % risk. → One position per engine, two engines.

- The "adaptive" stop that sometimes opens wide. Ninety-five trades lose 1 %, the next loses 8 %, the trade that breaks a prop-firm daily limit. → Sniper stop is $5, every trade; worst case is known before entry.

- Orders left in the market for days, picked off at 1 a.m. in the Asian session. → All pending orders withdrawn before daily close, fresh levels each session.

- Profit from news spikes, voided or flagged on funded accounts. → Ouro stands aside for news.

- "1 % max drawdown" from a fixed lot on a growing account — arithmetic, not safety. → % of balance sizing, relative equity drawdown published.

- 1-minute-bar backtests that overstate gold breakouts 3–4×. → Everything here is real ticks, plus the random-level test.

- No way to know what your broker is taking. → The Execution Quality panel.

IS IT CURVE FITTING?

I checked. The same engine, filters, period and pessimistic fills were run on 200 families of random price levels instead of the real ones. Real levels: +192 R. Random levels: 12 R on average, and the best of the 200 reached +98 R. That puts the real result 7.6 standard deviations above chance (p < 0.005). Not one of the 200 random versions came close. The levels carry the edge, not the tuning. You can check it yourselfs. 

REQUIREMENTS

- XAUUSD (or GOLD, depending on your broker), M5 chart, one chart per account.

- Hedging account. Each engine and the top-up are separate positions with their own stop, on a netting account they would merge.

- Low gold spread. Aim for 0.20 or less on average; above 0.30 you are handing the edge to your broker. This EA is sensitive to execution, and the panel will show you exactly how much.

- Suggested minimum: 2,000 at 1 % risk, or 1,000 at 2 %. Below that the partial close and the top-up cannot be sized and it is not the same EA. With 0.01 lots the smallest possible risk per trade is already 5 dollars. But of course you can start with lower capital. 

DON'T BUY THIS IF

You want daily trades. You trade a broker with wide gold spreads. You cannot leave it running for at least a quarter. You plan to use it on anything other than gold. A two-week flat stretch will make you turn it off. If none of that is you, welcome to the cave.

You can see both results and statistics of the EA in the images attached. Thank you for trusting the dragon!

I built this EA for my own account first and I keep running it. If something in it looks wrong to you, if you think a rule could be better, or if you simply do not understand why it did what it did on a given day, write to me. Private message here on MQL5, or the comments tab. I read everything and I answer.