Gold Session Breakout Pro EA
- Experts
-
Phan Van Khoa
⭐ PROFESSIONAL QUANT DEVELOPER & ALGORITHMIC TRADING SYSTEMS
Welcome to my official MQL5 developer page. I specialize in institutional-grade quantitative trading systems, high-precision Expert Advisors, and statistical breakout models for MetaTrader 5. - 버전: 4.20
- 활성화: 5
# Gold Session Breakout v420 PRO EA The session range is the market's opening bid — this EA takes it literally. On Gold M15 it measures the pre-session range, demands H4 EMA290 trend alignment plus RSI confirmation on both sides, and trades the buffered breakout with range-scaled stops and targets. Percentage-based trailing (start 150%, step 65%) lets runners extend in expansion regimes while the range multiple keeps risk proportional to the night's own volatility. ## Verified gate numbers (Model-0 real ticks, 7/7 PASS) | Window | Trades | PF | Net $ | DD eq% | Win% | |---|---|---|---|---|---| | Smoke 2024 | 72 | 1.24 | +265 | 2.48 | 37.5 | | IS 2021–23 | 197 | 1.26 | +862 | 4.32 | 39.1 | | OOS 2024–26 | 216 | 1.80 | +2432 (+24.32%) | 2.87 | 47.7 | | Full | 413 | 1.50 | +3289 (+32.89%) | 4.32 | 43.6 | Recovery factor 7.19. Monte Carlo (10,000 reshuffles): ruin probability 0.0%, 95th-percentile drawdown 6.14%. CPCV purged cross-validation: PASS — 100% of test splits positive. DSR 0.98: PASS. ## How it trades The EA measures the session range with a 190-point minimum (dead nights are skipped, not traded), adds a 25-point breakout buffer, and requires the H4 EMA290 trend plus RSI above 51 for longs and below 45 for shorts. Stops sit at 1.0× the range, targets at 2.1× — the multiple fixed, the distance adaptive. Trailing starts at 150% of entry distance in 65% steps, coarse enough to survive Gold's M15 noise. One trade per day maximum, Friday exit at 20:00, hard close 20:30. Frequency: 413 trades across five years, roughly 7 per month. A working sniper — active most weeks, quiet on dead-range nights by design. ## Honest limits Drawdown touched 4.32% in the IS window — inside the 5% gate but close enough to state plainly. The Monte Carlo 95th percentile reads 6.14%: in one reshuffle out of twenty, drawdown exceeds the gate cap. Size your deposit so 7% is survivable, not just 4.32%. Win rate 43.6% means the majority of individual trades lose; the 2.1× range multiple and trailing carry expectancy, and any regime that compresses follow-through (like 2021–23, PF 1.26) compresses returns toward single digits. The OOS years (PF 1.80) are the bull case, not the promise. Walk-forward honesty: 4-fold rolling WFO returned WFE 34.5% (FAIL against the 70% bar) with mean OOS net +$203 per fold. All four OOS windows were positive (+$12/+$202/+$487/+$109, min PF 1.03) — the edge is real across every fold — but in-sample captures roughly 3× what out-of-sample keeps. Expect decay from backtest numbers in live trading; the gate table above is the ceiling, the WFO mean (+$203/fold at 0.50% risk) the floor to plan around. ## Presets Default runs the gated configuration at 0.50% risk — the exact table numbers. Conservative halves risk to 0.25% for smaller accounts; expect roughly half the return at shallower drawdown. Balanced keeps 0.50% risk with identical entries (alternate magic for portfolio use alongside Default). Aggressive doubles to 1.00% — full-window drawdown scales toward 8%, suitable only with capital you can watch halve its cushion. ## Session timing and server hours Hours are server time (Exness MT5, UTC+0). The 190-point minimum range filters Asian-session dead zones; if your broker's GMT offset differs, verify the measured session still covers the pre-London lull before trading live — a misaligned window measures the wrong range and the multiples lose their premise. Spread cap and new-bar-only execution guard the entries; the 20:30 hard close avoids overnight gap exposure. ## Drawdown discipline The 4.32% IS print is the design constraint, not an accident. Every input that touches risk was set against it: the 190-point minimum range keeps position geometry out of dead zones where stops get harvested for nothing, the one-trade-per-day cap prevents revenge sequences, and the 1.5% daily-loss guard halts the session before a bad day compounds. Backtest the Conservative preset against 2021–23 if you doubt the margin — it holds under 2.5% there by construction. The Aggressive preset deliberately spends that margin for return; do not run it on accounts with external drawdown limits (prop-firm 5% daily caps will conflict with a 1.00%-risk session strategy on Gold). ## Version lineage v420 descends from the Gold_Session_Breakout line (v4 family gated through RND262–266, v5 shipped separately). The v420 calibration widens the H4 trend lookback to 290 and retunes the range multiples (1.0× SL, 2.1× TP) against the full 2021–26 window including the 2024–26 expansion regime — earlier calibrations were tuned on thinner OOS evidence. The Sep04 full re-run (RND435-class procedure) confirmed the numbers hold outside the original gate dates. Same-name earlier builds are not this binary: verify SHA `c0404388e203` before installing. Past performance does not guarantee future results. Trade at your own risk.
