Scalping Robot Pro MT5_2.0_fix… AI 测评｜QuantTester
Scalping Robot Pro MT5_2.0_fix · Me 报告：收益 37.2%、回撤 4.14%、PF 1.14、17236 笔交易。AI 风险解读。
策略安全信用评级: B (65 / 100) · 常态中性 (Moderate)
交易风险分析结果
Classified as Hybrid (Tiered Basket + Trend Following): a 641-day XAUUSD backtest produced 17,236 closed rows at 58.4% win rate and 1.14 profit factor (net $3,723.28 on $10,000) with a contained 4.14% maximum drawdown, but the edge is outlier-concentrated — stripping the top 5% of trades leaves a net loss of -$2,891.44 and PF 0.89 — making it a low-drawdown, thin-expectancy system whose live viability rests on a small number of large favorable excursions and on execution costs that the report does not model.
Darwinex 集中度压力测试 (Top 5% Windfall Dependence Test)
剔除排名前 862 笔（5%）极端盈利单后：毛利占比 22.49%，剩余净利润 $-2,891.44，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: $3,723.28
- 总收益率: 37.2%
- 最大回撤: 4.14%
- 总交易笔数: 17236
- 盈利因子 (PF): 1.14
AI 识别的核心高危漏洞清单
- [Return Consistency / R+] Outlier Dependency: The top 5% of trades (862 rows) account for 22.49% of gross profit; removing them leaves residual net profit at -$2,891.44 with PF 0.89, demonstrating that expectancy relies on a small set of large excursions rather than a broad repeatable edge.
- [Risk Stability / Rs] Loss Escalation Tail: stop_loss_evidence records max_over_median 9.83 with a largest single loss of -$47.74 and a maximum simultaneous exposure loss of -$76.24 across up to 5 orders, indicating that a minority of adverse sequences can erase many small wins and destabilize the loss distribution.
- [Payoff Asymmetry / La] With a 58.4% win rate and PF 1.14, average win is $2.92 versus average loss -$3.58 (payoff ratio 0.82); expectancy of $0.22 per trade is thin, so any widening of spread or slippage on XAUUSD at 4.1-minute average holding can flip arithmetic expectancy negative.
- [Execution Feasibility] The engine trades 26.88 times per day with average holding of 4.1 minutes on a scalping-class label; the backtest does not model commission, swap or live slippage, and 50-spread protection plus news filters imply the strategy already depends on favorable execution conditions that are unverified in live markets.
- [Momentum Decay / D-Score] Equity advanced from $10,000 to a peak near $13,700 without deep stagnation, but the March 2026 sequence shows clustered losses (-$29.09, -$23.67, -$16.82, -$12.57) whose recovery depended on subsequent large wins, illustrating that equity shape is sensitive to the timing of outlier trades.
客观数学打分明细
- [Base Score] All valid statements start at 50 points neutral benchmark (Grade B-)
- +5 pts [Order Basket]: Controlled multi-entry execution in same minute; layered basket architecture
- -15 pts [Windfall Reliance]: Removing top 5% outlier deals drops system into loss (Remaining PF 0.89 < 1.0)
- +15 pts [Minimal Drawdown]: Max historical drawdown only 4.1% (< 10%), excellent downside preservation
- +10 pts [Sample Horizon]: Sample duration spans 641 days (>= 180 days), covering diverse regimes
- Final Strategy Safety Score: 65 / 100 (Grade B).
下一步行动与优化建议
- Rerun the backtest with realistic broker commission and slippage applied to confirm the residual PF remains above 1.0.
- Export full trade rows with paired entry tickets and order comments to verify tiered-basket structure versus independent signals.
- Run a forward Demo test of at least 90-180 days and compare PF, average trade, and drawdown to the backtest baseline.
- Stress-test the top-5% exclusion on rolling windows to confirm whether other periods retain positive expectancy without outliers.
- Confirm whether the recovery/aggressive multipliers can activate in live configuration and cap their maximum exposure.
关于本报告的关键质询与答疑 (Q&A About This Report)
Why is the strategy rated Grade B rather than higher, given the 4.14% maximum drawdown?
The rating reflects outlier concentration rather than drawdown. Removing the top 5% of trades (862 rows) leaves residual net profit at -$2,891.44 and PF 0.89, meaning the system is profitable only when its best trades are included. The strong drawdown control (+15 pts) and the 641-day sample (+10 pts) support the Grade B, but the windfall reliance caps it below A-grade territory.
What is the realistic per-trade expectancy, and how fragile is it?
Expectancy is $0.22 per trade across 17,236 trades, with average win $2.92 against average loss -$3.58 (payoff ratio 0.82) and PF 1.14. At that margin, the strategy's profitability is highly sensitive to any unmodelled commission or slippage, which the report does not quantify.
Does the trade data show a grid or martingale pattern?
No. The parameters set maxOrders to 5, but the parsed rows show fixed 0.01-lot sizing throughout and no progressive lot escalation. Loss termination is clustered at a median of $2.96 with a tail extending to -$47.74, indicating risk is bounded by individual stop exits rather than compounding recovery sizing. The structure is consistent with a layered basket rather than a grid.
How should the stop-loss exits and the -$29.09 largest loss be interpreted?
The largest single loss is -$47.74, and stop_loss_evidence shows max_over_median of 9.83, meaning some losses run well beyond the median. This is consistent with a system that cuts most losses near a fixed level but occasionally permits a wider adverse excursion. The audit shows no stop-out or forced-liquidation event, so the risk is tail loss magnitude rather than account termination.
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.