# AuricFlow FTMO EA — Backtest Validation Results

**EA:** `AuricFlow_FTMO_EA.mq4` v1.00 · **Symbol:** XAUUSD M1 (every-tick model) · **Date:** 2026-06-12

## Test setup

- **Data:** real Dukascopy XAUUSD 1-minute candles, 2025-12-01 → 2026-06-11 (187,020 M1 bars, UTC), converted to MT4 `.hst`; synthetic tick volume ≈ 1 tick / 4 points of bar range (~47 ticks per M1 bar generated by the tester).
- **Terminal:** Capital Point Trading MT4 (build 1471), headless Strategy Tester, fixed spread **25 points** ($0.25) — plus a full stress rerun at **35 points**.
- **Deposit:** $10,000 (all FTMO limits are % of initial, so results map 1:1 to the $100k trial — read $ figures ×10).
- **FTMO objectives simulated:** target +5%, max daily loss 5% (CE(S)T-anchored), max total loss 10%, min 2 trading days.
- **11 rolling 14-day windows** (Jan 5 → Jun 8, overlapping starts every 2 weeks) + one **6-month run** per profile.

## Profile comparison (11 × 14-day windows each)

| Profile | Risk/Grid | Basket stop | Passes | Breaches | Median days to pass | Notes |
|---|---|---|---|---|---|---|
| v1 baseline | 2% | 1.5% | 0 / 11 | 0 | — | Safe but far too slow (+0.6%/window avg) |
| v2 push | 4% | 2.0% | 2 / 11 | 0 | 6.7 | More swing, edge diluted |
| v3 fast | 6% | 2.0% | 4 / 11 | 0 | 3.3 | Decent, deeper red windows |
| **v4 max (FINAL)** | **8%** | **2.5%** | **8 / 11** | **0** | **2.6** | Best pass rate, fails are soft |
| v5 refine | 8% | 2.0% | 5 / 11 | 0 | 2.5 | Tighter stop chops winners out |
| v4 @ 35pt spread | 8% | 2.5% | 8 / 11 | 0 | 2.6 | Spread-robust (same windows pass) |

## Final profile (v4) — window detail

| Window | Dates | Net (on $10k) | Result | Days to pass | Max daily budget used | Max equity DD |
|---|---|---|---|---|---|---|
| w01 | Jan 05–19 | +$555 | ✅ PASSED | 2.0 | 32% | 3.0% |
| w02 | Jan 19–Feb 02 | +$651 | ✅ PASSED | 2.0 | 13% | 2.5% |
| w03 | Feb 02–16 | −$497 | ✖ no pass | — | 50% | 5.1% |
| w04 | Feb 16–Mar 02 | −$597 | ✖ no pass | — | 47% | 6.4% |
| w05 | Mar 02–16 | +$564 | ✅ PASSED | 11.8 | 49% | 6.6% |
| w06 | Mar 16–30 | +$651 | ✅ PASSED | 2.5 | 30% | 2.9% |
| w07 | Mar 30–Apr 13 | +$568 | ✅ PASSED | 2.7 | 9% | 1.8% |
| w08 | Apr 13–27 | +$651 | ✅ PASSED | 8.7 | 43% | 4.5% |
| w09 | Apr 27–May 11 | +$523 | ✅ PASSED | 2.5 | 41% | 3.0% |
| w10 | May 11–25 | −$615 | ✖ no pass | — | 49% | 6.2% |
| w11 | May 25–Jun 08 | +$516 | ✅ PASSED | 3.2 | 49% | 3.8% |
| 6-month | Dec 08–Jun 08 | +$651 | ✅ PASSED | 4.4 | 53% | 4.4% |

**Pass rate 73% (8/11). Zero FTMO floor breaches in all 72 backtests across all profiles.**
The 3 failed windows ended at −5…−6% equity with the FTMO Guard's pre-trade wall blocking
further risk — i.e. the trial would simply expire un-passed, never breached. Max daily
budget usage never exceeded 53% of the $5,000 limit in the final profile.

## Guard verification (forced-breach tests)

| Test | Result |
|---|---|
| Daily hard stop (tiny 1.2% daily limit, basket stop off) | ✅ "daily loss hard buffer reached" → closed all at 80% of budget, halted until next CE(S)T day, `breached=no` — see [guardtest_breach_daily.htm](reports/guardtest_breach_daily.htm) |
| Total-loss floor (tiny 1.5% total limit) | ✅ "FTMO EMERGENCY: total loss buffer reached" → closed all, permanent halt — see [guardtest_breach_total.htm](reports/guardtest_breach_total.htm) |
| Profit-target lock (tiny 0.25% target) | ✅ Closed all at target, "CHALLENGE PASSED", token 0.01-lot trade auto-opened the next CE(S)T day → trading days 2/2, flat afterwards — see [guardtest_target_lock.htm](reports/guardtest_target_lock.htm) |
| Pre-trade worst-case wall | ✅ With limits smaller than one basket's worst case, the EA refused 100% of entries ("trade refused — worst-case basket loss would breach the daily/total buffer") |

## Reports

All MT4 Strategy Tester reports (HTML + equity-curve GIF) are in [reports/](reports/):
`w01_jan05_v4.htm` … `w11_may25_v4.htm`, `long_6mo_v4.htm` (final profile), the same set
for v1/v2/v3/v5/v4s35, the smoke test, and the three guard tests. Raw CSV summaries:
`results_v1.csv` … `results_v5.csv`, `results_v4s35.csv`.

## Honest caveats

- Backtests use bid-candle data with synthesized ticks (no real tick stream), fixed
  spread, and **no commission** — FTMO charges ~$3/lot/side on gold; with 3–13 trades
  per window the impact is roughly $20–$80 per window on $100k sizing, not decisive.
- 11 windows over 6 months of one regime (a strong, volatile gold market). A quiet,
  rangebound market would lower the pass rate (more windows ending like w03/w04/w10).
- The worst realistic outcome is bounded by design: the trial expires un-passed.
  A daily-loss breach would require a gap so large it blows through the basket stop
  AND the 20% safety buffer within one day — possible (e.g. a weekend-style gap), but
  the EA is flat before weekends and pauses around high-impact USD news.
- Past performance does not guarantee the next 14 days. ~70% pass odds is the honest
  read, not a promise.