Report analysis: GoldX9 strategy 9 · MT5 XAUUSD H1 Backtest
A long-horizon MT5 XAUUSD H1 backtest shows high absolute return (+174.69%) and an 86.55% win rate, but the 46.98% maximum drawdown, heavy dependence on a small number of large winning exits, and repeated equal-sized large losing clusters indicate asymmetric, cluster-driven risk that cannot be validated without paired-order/forward evidence.
This page analyzes a user-submitted MT4/MT5 report. It does not authenticate ownership, rerun an independent backtest or predict future profit.
The 0–100 AI score is an evidence-based assessment, not a probability of profit. Read it with the rationale and unknowns.
Read metric definitions, report comparisons and evidence limitsGoldX9 strategy 9 · MT5 XAUUSD H1 Backtest
The 0–100 AI score is an evidence-based assessment, not a probability of profit. Read it with the rationale and unknowns.
Data transparency
This page analyzes a user-submitted MT4/MT5 report. It does not authenticate ownership, rerun an independent backtest or predict future profit.
| Data source | User-submitted MT5 report |
|---|---|
| Broker | ForexTimeFXTM-Demo02 |
| Symbol | XAUUSD |
| Timeframe | H1 |
| Trade period | 2022-01-03 — 2026-01-29 |
| Report type | BACKTEST |
| Reported trades | 1279 |
| Processing method | Parser extraction and supported calculations + AI review; not account authentication |
Key metrics
These metrics come from imported records and supported calculations. Check capital, costs, cash flows and period before comparing reports.
| Return | 174.7% |
|---|---|
| Max drawdown | 46.98% |
| PF | 1.38 |
| Win rate | 86.55% |
| Net profit | 17,468.72 |
| Expectancy | 13.66 |
| Payoff ratio | 0.21 |
Growth
Normalised to 100 at the starting capital of the imported report. The chart samples the saved per-trade results and is not a live broker equity curve.
Normalised to 100 at the starting capital of the imported report. The chart samples the saved per-trade results and is not a live broker equity curve.
Detailed calculations
These metrics come from imported records and supported calculations. Check capital, costs, cash flows and period before comparing reports.
Return = net profit ÷ initial capital × 100%; PF = gross profit ÷ absolute gross loss; expectancy = net profit ÷ number of closed trades; payoff ratio = average win ÷ absolute average loss. Max drawdown uses the value reported by the source report. If the report contains deposits or withdrawals, the return should be interpreted together with the cash flow.
Why this safety score
Reported return is +174.69% on a 10,000 base (net 17,468.72) across 1,279 parsed closed rows, with profit factor 1.38 and win rate 86.55%.
- Maximum drawdown is 46.98%, which is large relative to a 1.38 profit factor and sits just under the 50% hard-loss band; balance ranged from a 10,000 start to a peak above 28,000 and down to roughly 8,000 in the visible rows.
- Profit concentration is material: computed stats show the top 1% of trades (13) produce 8.97% of gross profit and removing them drops profit factor to 1.25; the top 5% (64 trades) produce 30.64% of gross profit and removing them leaves net profit of -2,016.25 and profit factor 0.96.
- Many losses appear as same-close-timestamp groups of identical P/L and identical lot size (e.g., three -448.89 rows on 2022.01.12, four -1,135.40 rows on 2025.10.09, eight multi-row clusters in 2024.09), and the panel carries no entry tickets; pairing cannot be confirmed from the supplied rows.
- Trailing-stop style exits are visible: several stop-loss-type exits still show positive profit, which is a normal profit-lock mechanism rather than a data defect.
- Sample period is long (2022-01-03 to 2026-01-29, ~1,487 days) and History Quality is reported at 99%, but this measures the modelled history only.
- No commission, swap, fee, or slippage values are present in report_context; cost impact cannot be assessed from the supplied data.
Key risk flags
46.98% maximum drawdown against a 1.38 profit factor: a small edge is carrying very large equity swings.
- Profit is concentrated: removing the top 5% of trades turns the strategy net loss-making (remaining net -2,016.25, PF 0.96).
- Multiple identical-loss clusters close on the same timestamp with identical lot sizes (e.g., three -448.89 rows, four -891.47 rows, three -1,135.40 rows), which points to correlated exposure whose structure cannot be verified from the supplied rows.
- Individual losing exits reach roughly -1,342 (2026-01-29), -1,135 (2025-10-09 and 2026-01-16), and -973 (2025-06-17), dwarfing the typical small winning exit near +5 to +15, evidencing an unfavourable payoff asymmetry.
- Historical drawdown episodes in the visible rows repeatedly cut the balance by roughly 40-45% (e.g., from about 17,600 to about 8,000-10,000 in 2024.09 and from about 28,000 to about 19,500 in 2025.10).
What the data cannot tell
Entry timestamps, entry tickets, and trade direction on entry are not in the supplied rows, so simultaneous openings, grid/scale-in structure, and basket correlation cannot be confirmed.
- Commission, swap, spread, and slippage settings are absent, so the cost-adjusted edge is unknown.
- Whether the repeated equal-P/L, same-timestamp exits are independent trades or components of a multi-position structure is unresolved.
- Behaviour in live or forward accounts, including real fills, slippage, and broker minimum-holding rules, is not observable from this report.
Next actions
Re-export the trade list with entry tickets and open timestamps so same-timestamp clusters can be paired and classified.
- Recompute performance after removing the top 5% and top 1% of trade P/L to quantify dependence on large winners.
- Run the same model with explicit commission, swap, and slippage values representative of the intended broker to test cost sensitivity.
- Reconcile the gap between the 2026-10-05 report range end and the 2026-01-29 last parsed close, and confirm no trades are missing.
- Forward-test on a Demo account for a defined period and compare fill quality and drawdown against the backtest.
Frequently asked questions
The 0–100 AI score is an evidence-based assessment, not a probability of profit. Read it with the rationale and unknowns.
Q. Why is the maximum drawdown of 46.98% a concern when profit factor is 1.38?
A. The report pairs a relatively modest 1.38 profit factor with a 46.98% maximum drawdown on a 10,000 base; the visible balance path falls from over 28,000 toward 19,500 in late 2025 and from roughly 17,600 toward 8,000 in 2024, meaning the equity swings are large relative to the edge.
Q. How dependent are the results on a few trades?
A. The computed stats show the top 1% of trades (13) contribute 8.97% of gross profit (remaining PF 1.25) and the top 5% (64) contribute 30.64%; removing the top 5% leaves net profit at -2,016.25 with profit factor 0.96.
Q. The rows show many profitable 'out' events with s/l-style exits; is that a data problem?
A. No, this is consistent with the strategy using a trailing or breakeven stop that locks in profit before price retraces; it is a normal exit style, not a data defect, and it explains the 86.55% win rate alongside the small average winning exit size.
Q. What does the 99% History Quality and the long 2022-2026 window tell us?
A. The report states 99% History Quality with 27,464 H1 bars and 6,535,550 ticks over 2022.01.03-2026.10.05, and parsed closed trades span 2022-01-03 to 2026-01-29 (about 1,487 days), so the sample is long and well-modelled; this measures the backtest history only and does not validate live fills or the cost model, which contain no commission or slippage values in the supplied report context.
Q. How is RBOT different from asking a general-purpose AI assistant (such as ChatGPT, Gemini, Claude or DeepSeek) to analyse my report myself?
A. RBOT parses MT4/MT5 reports into structured metrics and trade records before AI review. Supported calculations can be checked against the original export; AI wording and scores can vary and are not guaranteed to be deterministic or comparable across reports. General AI assistants may also use calculation tools. Missing actual slippage, ownership or forward validation still requires other evidence.