GOLD GPT EA M5_fix · MetaTrader 5… AI Review | QuantTester
GOLD GPT EA M5_fix · MetaTra: return 26.7%, DD 1.43%, PF 1.43, 2953 trades. Read the review.
Strategy Safety Grade: A (80 / 100) · Robust & Controlled
Trading risk analysis result
Strategy Type: Strict Lot Discipline. Classified as Strict Lot Discipline (fixed 0.01-lot single-order execution) on XAUUSD M5: the strategy generated 26.69% net return at a 1.43 profit factor with a maximum equity drawdown of only 1.43%, a Calmar-style return/drawdown ratio near 18.7; the structural weak point is payoff asymmetry (average win 3.85 versus average loss -9.31, ratio 0.41) that is mathematically offset by a 77.5% win rate, while realized-loss clustering at -11.00 (median = max = 11.00, 80.9% clustered) evidences disciplined per-trade stop termination.
Darwinex 集中度压力测试 (Top 5% Windfall Dependence Test)
剔除排名前 148 笔（5%）极端盈利单后：毛利占比 22.62%，剩余净利润 $675.31，剩余 PF 1.11。
Data transparency & Environment
- MT4/MT5: MetaTrader 5
- Symbol: XAUUSD
- Timeframe: M5 (2025.01.03 - 202
- Net profit: $2,668.64
- Return: 26.7%
- Max drawdown: 1.43%
- Total Trades: 2953
- Profit Factor (PF): 1.43
Critical Algorithmic Vulnerabilities
- [Return Consistency Defect / R+ Outlier Dependency] The top 1% of trades (30 rows) carry 7.78% of gross profit and the top 5% (148 rows) carry 22.62%; removing the top 5% leaves residual net profit of only 675.31 at PF 1.11, meaning the vast majority of the 26.69% return depends on a small tail of outsized winners — in live copy-trading, a regime where those fat winners fail to materialize compresses returns toward break-even.
- [Risk Stability / Rs] Loss scale is capped but not scalar-stable across regimes: 80.9% of the 455 analyzed losing trades cluster at exactly -11.00 with max/median = 1.00, evidencing a hard per-trade stop, yet because average loss (-9.31) is 2.4x average win (3.85), any decay in the 77.5% win rate — for example to the mid-60s — inverts expectancy before the drawdown statistic would register it.
- [Payoff Asymmetry / La Loss Aversion] With a 0.41 payoff ratio, the arithmetic of the strategy requires a sustained high hit-rate; the longest losing streak observed is 4, but a clustered run of 8–10 consecutive -11.00 stops (approximately -11% of gross profit per 10 stops) would erase roughly two weeks of accumulated small wins given 4.58 trades per day.
- [Execution Feasibility] The 36.8-minute average holding period on XAUUSD M5 with average wins of 3.85 USD per 0.01 lot is a thin-margin profile; the supplied file carries no spread, commission, or slippage data, so the modest realized per-trade edge cannot be assumed to survive widened gold spreads or execution latency in live brokerage.
- [Momentum Decay / D-Score] The equity trajectory increments gradually (balance 10,000 to 12,669.78) but the gross-profit contribution is front-loaded into a handful of large single-day bursts (e.g. 61.87 on 2026-01-30, 41.08 on 2025-10-29, 39.75 on 2026-07-02); dormant periods between these bursts produce a flat quote shape with episodic returns rather than steady alpha.
Score Breakdown Details
- [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
- -20 pts [Poor Payoff]: Win/loss payoff ratio only 0.41 (< 0.6), losses heavily erode gains
- +5 pts [Expectancy]: Overall Profit Factor reached PF 1.43 with solid positive mathematical expectancy
- +5 pts [Organic Break-even]: Removing top 5% windfall deals leaves PF 1.11 >= 1.0; maintains break-even organically
- +15 pts [Minimal Drawdown]: Max historical drawdown only 1.4% (< 10%), excellent downside preservation
- +10 pts [Sample Horizon]: Sample duration spans 645 days (>= 180 days), covering diverse regimes
- Final Strategy Safety Score: 80 / 100 (Grade A).
Recommended Next Actions
- Run a forward Demo or micro-live test on the same XAUUSD broker feed to quantify realized spread and slippage erosion against the 3.85 USD average win.
- Stress-test the 77.5% win rate by re-simulating on a higher-spread cost model to confirm expectancy remains positive after friction.
- Verify paired entry/exit tickets from the raw MT5 statement to confirm the maximum concurrent-exposure profile and rule out undetected correlated baskets.
- Segment performance by calendar year and volatility regime to confirm that the top-5% profit contribution is not concentrated in a single quarter.
- Evaluate sensitivity of the 1.43% maximum drawdown to a 2x spread scenario, since a drawdown statistic that small is the primary allocator selling point.
Q&A About This Report (Quant Due Diligence)
The strategy achieved a Grade A Safety Grade; what are the core strengths and rationale supporting this rating?
The Grade A 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.
Is the 26.69% return driven by a handful of lucky trades?
Partly, but not exclusively. The top 5% of trades (148 rows) account for 22.62% of gross profit; excluding them leaves residual net profit of 675.31 at PF 1.11, still above breakeven. The top 1% (30 rows) accounts for 7.78% of gross profit. So the return is tail-influenced but the underlying system retains a modest organic edge without those outliers.
Is there evidence of martingale or grid scaling?
No. All parsed rows execute at a constant 0.01 lot with audit facts confirming martingale_stepup = false, zero stop-outs, and a maximum of 2 simultaneous orders (max simulated loss 11.87). The realized-loss distribution clusters tightly at -11.00 (median = max = 11.00, 80.9% clustered share), which is the signature of a fixed per-trade stop rather than a compounding adverse scale-in.
How reliable is the drawdown of 1.43% as a live risk estimate?
It is the maximum equity drawdown recorded over 644.7 days of backtest at a 10,000 capital base with 26.69% net return, giving a return-to-drawdown ratio of approximately 18.7. That figure is internally consistent with the tight -11.00 loss clustering, but the supplied report carries no spread, commission, or slippage fields, so the live drawdown may widen if gold execution costs erode the average win of 3.85 USD per 0.01 lot.
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.