RBOT EVIDENCE LAB
公开 AI 交易测评

An MT5 XAUUSD M1 backtest (2025-01-06 to 2026-04-24, 588 closed rows) reports +12,792.9% on a 1,000 base with PF 5.07 and 11.33% max drawdown, but the sample is trade-row-incomplete and lacks paired entry tickets, so the profit concentration and true risk structure cannot be fully verified.

分析时间:2026-10-06 17:05
回测

MT5 XAUUSD M1 Backtest

62/100策略安全评分
12,792.9%收益
11.33%最大回撤
5.07PF
588交易数

数据透明度

数据来源用户上传的 MT5 报告
品种XAUUSD
周期M1
报告类型回测
已核算交易588
核验方式RBOT 解析器逐笔核算 + AI 风险审计

核心指标

收益12,792.9%
最大回撤11.33%
PF5.07
胜率70.75%
净利润127,929.04
单笔期望217.57
平均盈亏比2.09

增长

以导入报告的起始资金归一化为 100。该图按已保存的逐笔结果采样,不等同于经纪商实时权益曲线。

详细计算结果

1,000.00初始资金
127,929.04净利润
70.75%胜率
159,394.75总盈利
31,465.71总亏损(绝对值)
383.16平均盈利单
-182.94平均亏损单
2.09平均盈亏比
217.57单笔期望
3,500.00最大盈利单
-3,500.00最大亏损单
23最长连胜
5最长连亏
0.50平均手数
473.4 天样本跨度
1.24日均交易

收益率 = 净利润 ÷ 初始资金 × 100%;PF = 总盈利 ÷ 总亏损绝对值;单笔期望 = 净利润 ÷ 已平仓交易数;平均盈亏比 = 平均盈利单 ÷ 平均亏损单绝对值;最大回撤采用报告输出的最大回撤值。若报告含入金或出金,收益率应结合现金流另行解释。

策略安全评分依据

  • 588 parsed closed rows over 473.4 days with a 70.75% win rate and profit factor of 5.07 on an initial deposit of 1,000.
  • Reported maximum drawdown of 11.33% against a balance that ran from 1,000 to 128,929.04, giving a high return-to-drawdown ratio.
  • History Quality reported at 99% with no explicit tester warnings in evidence_flags, supporting execution assumptions for this run.
  • The top 1% of trades (6 rows) account for 13.17% of gross profit; the top 5% (30 rows) account for 41.19%, leaving a residual profit factor of 2.98, so the edge is not entirely dependent on a tiny cluster of rows.
  • The top-5% concentration (41.19% of gross profit) and the reported fixed SL/TP of 3500.0 mean the loss tail and recovery behaviour depend on a small set of outcomes that the capped trade-row sample does not fully expose.
  • Position size rises from 0.03 lots early to 1.0 lots later, consistent with balance-proportional compounding rather than a verified escalation rule; the report does not disclose the sizing formula.
  • Multiple -3500.0 rows appear (e.g. 2025-12-24, 2026-01-30, and a +3500.0 cluster on 2026-01-29), indicating full-size stop and target events at the configured fixed SL/TP; these are closed events and cannot be assumed to be grid or basket exits without entry-ticket pairing.

关键风险信号

  • Profit concentration: the top 5% of closed rows produce 41.19% of gross profit, and the top 1% produce 13.17%; the residual profit factor after removing the top 5% falls to 2.98 from 5.07.
  • Loss tail: individual closed rows show -3500.0 at full size (2025-12-24, 2026-01-30), matching the configured fixed stop; with position size at 1.0 lot and a 1,000 base, these single outcomes are large relative to early-balance equity.
  • Maximum drawdown of 11.33% is a single reported summary statistic computed on the running balance; its timing and duration relative to the 473.4-day trade window cannot be located without an equity curve or paired position data.
  • The closed-row sample does not expose open exposure at any point, so overlapping or simultaneous positions behind repeated close timestamps (e.g. the 2026-01-08 cluster and 2024-04-16 cluster) remain unverified.
  • Position size increases with balance (0.03 to 1.0 lots) and no explicit sizing rule is documented, so the relationship between balance growth and per-trade risk is inferred, not verified.

现有数据无法判断

  • Entry timestamps, ticket numbers, and paired entry-to-exit mapping are absent from the supplied rows, so holding times, overlap, and position structure cannot be determined.
  • Whether the repeated close timestamps (e.g. the 2026-01-08 sequence and 2024-04-16 sequence) represent independent positions, scale-ins, or grid-type management cannot be established without paired-ticket evidence.
  • The exact position-sizing rule and how it reacts to drawdown are not documented.
  • Whether the reported 11.33% maximum drawdown reflects the same position-level risk seen at 1.0-lot size late in the sample cannot be confirmed from summary statistics alone.
  • Real-world slippage and commission impact are not modelled in the supplied fields and cannot be quantified from this report.

下一步行动

  • Re-export or re-parse the closed-row data with entry ticket, entry time, and paired order identifiers to verify overlap, holding periods, and whether repeated close timestamps represent independent or grouped positions.
  • Recompute maximum drawdown and recovery from a dated equity curve, including the 2025-12 to 2026-01 loss cluster, rather than relying on the single summary 11.33% figure.
  • Document the position-sizing rule and recompute risk per trade at 1.0-lot size relative to the contemporaneous balance to confirm whether relative risk rises late in the sample.
  • Run a cost-sensitivity check with a realistic spread/commission assumption for XAUUSD, since no cost fields appear in report_context.
  • If forward validation is intended, run the same logic on a Demo or Real account and compare trade-level outcomes with this Backtest before drawing any live-performance conclusion.

常见问题

Q. The report shows a 12,792.9% return on a 1,000 base with only an 11.33% maximum drawdown; how much of that return comes from a small number of trades?

A. The computed trade statistics show that the top 1% of closed rows (6 rows) account for 13.17% of gross profit and the top 5% (30 rows) account for 41.19%. After removing the top 5%, the residual profit factor is still 2.98, so the result is not built on a single trade, but it does rely on a concentrated set of larger winners.

Q. Several rows close at +3500.0 and -3500.0; does that indicate grid or basket trading?

A. Those values match the configured fixed SL of 3500.0 and fixed TP of 3500.0 in report_context, so they are consistent with individual stop-loss and take-profit exits at full position size. Because the supplied rows contain only close timestamps, direction, size, and profit with no paired entry tickets, they cannot be described as grid, basket, or multi-position exits from this evidence.

Q. The trade sample runs from 2025-01-06 to 2026-04-24 but the chart period ends 2026.10.05; why is there a gap?

A. The full_trade_period is calculated from the 588 parsed closed rows, which end on 2026-04-24, while the report_context chart range extends to 2026.10.05. The gap reflects the trade-row window rather than an error; entry timestamps are not supplied, so the reason no closed rows are shown after 2026-04-24 cannot be determined from this data.

Q. How reliable is this backtest given the reported test settings and the absence of forward records?

A. History Quality is reported at 99%, with 616,672 bars and 2,466,687 ticks over the M1 XAUUSD range, and no tester warnings appear in evidence_flags. However, entry tickets, entry times, commission, swap, and slippage fields are not present, and no Demo or Real forward record is supplied, so the execution assumptions behind the 70.75% win rate and 5.07 profit factor remain unverified.

Q. 自己把报告丢给通用 AI 分析,和用 RBOT 分析有什么不同?

A. 通用 AI 只能看到你粘贴的文本或截图,所有数字都得它自己算 —— 幻觉往往就出在这里。RBOT 先用专为 MT4/MT5 写的解析器把报告解析成结构化结果,再交给固定的审计提示词评审,所以安全评分口径统一、可复现、也能在不同报告之间横向比较。每个指标的计算过程都公布在本页,你可以自己复核;对报告本身无法证明的数据(真实滑点、点差、前向验证)RBOT 也不下结论,只列为待核验。

RBOT.CC · EA 与量化交易系统平台使用 RBOT 测评工具查看全部公开测评