SuperBot V4 · MT5 XAUUSD H1 Backtest AI 测评｜QuantTester
SuperBot V4 · MT5 XAUUSD H1 Backtes 报告：收益 65.0%、回撤 14.78%、PF 1.31、359 笔交易。AI 风险解读。
策略安全信用评级: A+ (90 / 100) · 稳健受控 (Robust)
交易风险分析结果
Strategy Type: Multi-Strategy Portfolio. Classified as Multi-Strategy Portfolio: the SuperBot V4 backtest on XAUUSD H1 (2025-01 to 2026-10, 359 closed rows) produced a 65.0% return with a 1.31 profit factor, 57.94% win rate and a contained 14.78% maximum drawdown (Calmar ~4.4), retaining PF 1.16 after removing the top 5% of trades — a structurally sound, only-moderately outlier-dependent book whose principal residual exposure is the flat 2000-point stop with an essentially 1:1 realized payoff (0.95) and the single-symbol H1 execution-friction assumption set.
Darwinex 集中度压力测试 (Top 5% Windfall Dependence Test)
剔除排名前 18 笔（5%）极端盈利单后：毛利占比 11.41%，剩余净利润 $3,380.00，剩余 PF 1.16。
数据透明度与环境说明
- MT4/MT5: MT5
- 品种: XAUUSD
- 周期: H1
- 总净利润: $6,500.00
- 总收益率: 65.0%
- 最大回撤: 14.78%
- 总交易笔数: 359
- 盈利因子 (PF): 1.31
AI 识别的核心高危漏洞清单
- [Return Consistency / Outlier Dependency] Profit is moderately concentrated rather than structurally dependent: the top 1% of trades (4 rows) contribute 2.63% of gross profit and the top 5% (18 rows) contribute 11.41%; removing the top 5% still leaves net 3,380 and PF 1.16. The asset-management hazard is therefore low — the edge is not a small-sample windfall — but the modest 11.41% concentration means per-trade returns remain thin and sensitive to execution cost drift.
- [Risk Stability / Rs] Stop and target distances are near-symmetric: average win 131.44 versus average loss -138.01 (realized payoff 0.95), largest win 180 versus largest loss -180, and single losses of -160 to -180 recur throughout 2026 (e.g. 2026-06-24 -180, 2026-07-16 -180, 2026-07-20 -180). Because win-rate (57.94%) carries the expectancy, a modest rise in stop-out frequency or any widening of effective spreads on XAUUSD would compress the margin between the 0.95 payoff and the ~1.31 PF.
- [Risk Stability / Rs] Dollar risk scales with balance through lot graduation (0.04-0.05 lots in early 2025, 0.06-0.07 mid-2025, 0.08-0.09 by mid-2026), so decline-phase losses at the 0.09 tier (-180 per event) are roughly 2.25x the early-phase -80 losses in absolute dollars. This is proportionality to equity, not escalation of relative risk, but it means a later-period drawdown converts to a larger balance percentage than the same losing streak would have in 2025.
- [Momentum Decay / D-Score] The equity curve's strongest ascent occurred in the first half of 2026 (balance 15,720 in early January to 18,900 in mid-July), after which a choppy stretch (e.g. 2026-08-25 to 2026-09-25 with repeated -160 exits) brought balance back to 16,340-16,660 before partial recovery to 16,500 at the 2026-10-09 close. The 14.78% maximum drawdown sits inside this later consolidation, indicating intermittent alpha stagnation that must be monitored for persistence rather than treated as a one-off.
- [Execution Feasibility / Capacity] The backtest is an idealized H1 tester run on a single symbol with a 2000-point fixed stop and no verifiable live broker fills; because average winners (131.44) and losers (-138.01) are the same order of magnitude, live slippage and spread widening on gold directly alter the 1.31 PF. There is no verified forward or live record to corroborate that the tester-assumed fills are attainable.
客观数学打分明细
- 【Baseline】Starts at 50 pts neutral baseline
- +15 pts / Disciplined Single Order: Fixed or proportional lot size with strictly disciplined execution, zero counter-trend averaging down
- +5 pts / Verified Positive Profit Factor: Full-sample PF is 1.31 (net profit 6500.00), demonstrating fundamental positive expectancy
- +5 pts / Resilient Baseline Post-Top 5%: Retains PF 1.16 >= 1.0 (net 3380.00) after removing top 5% trades, maintaining capital preservation baseline
- +5 pts / Controlled Drawdown (10%-20%)
- +10 pts / Superior Calmar Ratio: Return-to-drawdown ratio is 4.4 (>= 2.5), generating substantial edge per unit of downside risk
- Final Strategy Safety Score: 90 pts (Grade A+).
下一步行动与优化建议
- Retain the full MT5 report (with entry timestamps, ticket pairing and per-trade SL/TP prices) to verify whether the 0.04-0.09 lot tiers follow a documented balance-proportional rule and whether any intra-trade exposure was concurrent.
- Re-run the same 2025-01 to 2026-10 window with commission and a realistic XAUUSD spread/slippage assumption (e.g. 20-40 points) to test the durability of the 0.95 payoff and 1.31 PF.
- Isolate the post-2026-07 equity segment and quantify the drawdown within it, comparing trade frequency and win rate against the first half of 2026 to distinguish regime chop from structural decay.
- Escalate from backtest to a forward Demo run on the intended broker feed for at least one quarter, capturing realized fills, then compare PF, average win/loss and trade frequency against the backtest baseline.
- Add a per-trade risk-budget check around the fixed 2000-point stop to confirm that at the 0.09 lot tier the -180 loss per event remains within the intended percentage-of-equity risk tolerance.
关于本报告的关键质询与答疑 (Q&A About This Report)
With a 57.94% win rate but a realized payoff ratio of only 0.95, what is the strategy's actual source of expectancy?
Expectancy is win-rate-driven, not payoff-driven. Gross profit of 27,340 across 359 closed rows against gross loss of 20,840 yields a profit factor of 1.31 and a per-trade expectancy of 18.11. The 57.94% hit rate covers the near-symmetric payoff (average win 131.44, average loss -138.01). Because the payoff margin is thin, the strategy is sensitive to any drift in win rate or in execution cost, which is why the 95% post-top-5% PF of 1.16 is a meaningful resilience floor rather than a large cushion.
Is the 65% return simply a product of a handful of outsized trades?
No, the concentration is modest. The top 1% of trades (4 rows) account for only 2.63% of gross profit and the top 5% (18 rows) for 11.41%; stripping out the top 5% still leaves a residual net profit of 3,380 with a PF of 1.16. The largest single win (180) equals the largest single loss (-180), and no individual trade dominates the distribution. The edge therefore appears systematic and repeatable across the 640-day sample rather than an outlier windfall.
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.
The equity curve pulled back from an 18,900 peak to a 16,340 trough — does the later consolidation undermine the backtest result?
It represents a monitored, contained phase rather than a breakdown. Balance rose from 15,720 in early January 2026 to 18,900 on 2026-07-13, then moved in a choppy band (repeated -160 exits from 2026-08-25 through 2026-09-25) before closing at 16,500 on 2026-10-09. This decline is fully consistent with the reported 14.78% maximum drawdown and did not break the trade-level discipline — all exits are within the -160 to -180 bound. The principal open question is whether this consolidation persists, which cannot be resolved from the single backtest and requires forward observation.
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.