Green Profit gold ea · MetaTrader 5… AI 测评｜QuantTester
Green Profit gold ea · MetaTrader 5 报告：收益 55.2%、回撤 11.27%、PF 1.54、204 笔交易。AI 风险解读。
策略安全信用评级: B+ (70 / 100) · 常态中性 (Moderate)
交易风险分析结果
Strategy Type: Strict Lot Discipline. Classified as Single-Order Trend (discretionary ATR/EMA-filtered swing system with auto-lot sizing): the 204-trade, 425-day XAUUSD backtest produced a 55.21% net return at a 1.54 profit factor with a contained 11.90% maximum drawdown, but expectancy is fat-tail dependent—removing the top 5% of trades leaves a 0.89 residual PF—so this is a viable but outlier-concentrated B+ system whose forward edge hinges on capturing its infrequent trend runs.
Darwinex 集中度压力测试 (Top 5% Windfall Dependence Test)
剔除排名前 11 笔（5%）极端盈利单后：毛利占比 42.21%，剩余净利润 $-1,135.35，剩余 PF 0.89。
数据透明度与环境说明
- MT4/MT5: MetaTrader 5
- 品种: XAUUSD
- 周期: H1 (2025.01.03 - 202
- 总净利润: $5,520.57
- 总收益率: 55.2%
- 最大回撤: 11.27%
- 总交易笔数: 204
- 盈利因子 (PF): 1.54
AI 识别的核心高危漏洞清单
- Outlier Dependency (Return Consistency Defect / R+): The top 5% of trades (11 rows) account for 42.21% of gross profit; removing them leaves residual net profit at -$1,135.35 with a PF of 0.89, demonstrating that expectancy is driven by statistical outliers rather than a durable systematic edge.
- Payoff Asymmetry Fragility (La Loss Aversion): With a 47.06% win rate, payoff ratio 1.73 and average win $164.24 vs average loss -$94.87, arithmetic expectancy depends on occasional outsized winners (largest win $861.40 vs largest loss -$155.78); a prolonged choppy regime that truncates trend expansion would flip the residual arithmetic negative (post-top-5% PF 0.89).
- Risk Stability Breakdown (Rs): Loss scale drifted upward through the sequence—early losses clustered around -$22 to -$104 (e.g. -$22.55 on 2025-01-29, -$10.35 on 2025-02-06) versus late-sequence cluster of -$123 to -$156 (e.g. -$143.73 on 2026-01-13, -$155.78 on 2026-02-13)—indicating expanded dollar exposure via auto-lot scaling rather than a stable fixed risk unit.
- Execution Friction Sensitivity (Capacity & Execution Feasibility): H1 gold entries with average hold of 1,896.6 minutes (≈31.6 hours) mean the edge is trend-duration-driven; however the auto-lot mechanism (1% risk, max 1.0 lot) and fixed 55,555-point TP / 25,555-point SL parameters imply wide nominal stops whose realised terminal losses (~-$90 to -$156) depend on unspecified broker cost and fill assumptions that cannot be verified from this report.
客观数学打分明细
- [Base Score] All valid statements start at 50 points neutral benchmark (Grade B-)
- +15 pts [Lot Discipline]: Fixed/proportional lot execution, no adverse counter-trend doubling
- +5 pts [Expectancy]: Overall Profit Factor reached PF 1.54 with solid positive mathematical expectancy
- -15 pts [Windfall Reliance]: Removing top 5% outlier deals drops system into loss (Remaining PF 0.89 < 1.0)
- +5 pts [Controlled Drawdown]: Max historical drawdown is 11.9% (10%-20%), within stable controlled zone
- +10 pts [Sample Horizon]: Sample duration spans 425 days (>= 180 days), covering diverse regimes
- Final Strategy Safety Score: 70 / 100 (Grade B+).
下一步行动与优化建议
- Obtain a paired entry/exit trade journal (ticket-linked) to confirm whether exits were signal, trailing, or hard-stop executions, and reconcile zero SL/TP fields against InpFixedSLPoints/InpFixedTPPoints.
- Rerun the backtest with explicit realistic commission, swap, and variable spread to quantify cost sensitivity on the 1.54 PF and the 0.89 post-top-5% residual.
- Run a Monte Carlo or bootstrap on the 204-trade P/L distribution to estimate the probability that a forward window fails to capture its trend-run trades and slips below breakeven.
- Test on a second gold dataset or broker feed for cross-feed robustness of the ATR/EMA cloud filter (InpCloudLength=56, InpEMAPeriod=200).
- Conduct a forward paper-trading phase over at least 3-6 months to validate auto-lot equity scaling and live slippage behaviour on H1 gold.
关于本报告的关键质询与答疑 (Q&A About This Report)
The strategy achieved a Grade B+ Safety Grade; what are the core strengths and rationale supporting this rating?
The Grade B+ rating is backed by robust risk controls in the trade data: strict execution discipline (e.g. pre-set stop losses, synchronized basket entries without toxic martingale averaging), and historical drawdowns kept within institutional thresholds.
The strategy wins only 47.06% of trades—how does it still produce a positive 1.54 profit factor?
Through payoff asymmetry: average winners are $164.24 versus average losers of -$94.87 (payoff ratio 1.73), and the largest single win of $861.40 dwarfs the largest loss of -$155.78. Expectancy is positive per trade at $27.06, but it is concentrated—removing the 11 best trades turns net profit negative (-$1,135.35 residual).
The trade rows show zero SL and TP with identical entry and close prices—does this mean the EA runs unprotected?
No such conclusion can be drawn from the exported rows. The parameters do declare stop/target inputs, and the realised losing trades do terminate in a common -$90 to -$156 band (median -$107.92), consistent with a controlled exit level rather than an unbounded drawdown. However, because the export does not carry populated SL/TP or paired entry data, the exact execution mechanism—signal exit, trailing stop, or fixed stop—cannot be confirmed from this file and should be verified against a ticket-paired journal.
The late-sequence losses (-$123 to -$156) are larger than the early ones—is risk inflating over time?
Yes, that pattern is present: early losses clustered near -$10 to -$104, while from roughly late 2025 the losing cluster widened to -$123 to -$156. This is consistent with auto-lot sizing that scales exposure as the $10,000 balance grew to $15,520.57 by 2026-03-04 rather than with adverse counter-trend averaging, so it is recorded as dynamic exposure growth, not a martingale signature; nonetheless, dollar-loss expansion is a replication-stability (Rs) consideration for live allocation.
How is QuantTester different from asking a general-purpose AI assistant myself?
QuantTester parses MT4/MT5 reports into structured metrics and trade records before AI review. Supported calculations can be checked against the original export, guaranteeing consistent audit facts.
Which broker is recommended for trading and running MT4/MT5 strategies?
For algorithmic trading, EA execution and reliable backtesting, we strongly recommend IC Markets (Raw Spread Account) with 0.0 pip spreads.