LowDD Macro Pullback EA
Macro-gated multi-timeframe forex pullback
Rejected — the risk controls are careful, but the M5 signal lost 4.67% with PF 0.44 after modeled costs. Four predeclared variants also lost, and the least-bad hourly replacement failed its untouched test.
- Category
- Trend
- Window
- M5: Apr–Jul 2026; H1 controls: Aug 2024–Jul 2026
- Instruments
- EURUSD
- Timeframe
- D1 / M15 / M5, with H1 alternatives
- Tested
- 2026-07-19
Pending validation
A scaled equity curve appears here once this strategy clears the data needed to compute one honestly. We don't show a curve we can't stand behind.
See what the gates require →How it works
The bet
What market behavior this strategy is wagering on.
It bets that an RSI recovery after a shallow M5 pullback will resume a daily and intraday EURUSD trend, while explicit macro direction, volatility, spread, session and loss gates screen out the worst conditions.
How it decides
What makes it enter, size, and exit a position.
The EA aligns daily and M15 exponential moving averages, waits for price to pull into the M5 fast EMA, and enters when RSI crosses its recovery threshold with ADX confirmation. It sizes each trade to 0.25% equity and uses an ATR stop, fixed reward target, break-even rule and daily circuit breakers.
How it can break
The regimes and failure modes that turn the edge negative.
The entry signal does not overcome execution costs or failed continuation. Its 46.9% win rate looks plausible in isolation, but average outcomes after the modeled 2.1-pip round trip produce −0.389R expectancy and losses in development, validation and untouched-test segments.
Explainer compiled 2026-07-19 · codex-clean-room-v1.1
Market context
Live chart
Chart powered by TradingView. Live prices can differ from the point-in-time dataset used in the published test.
The story in one line
LowDD Macro Pullback is a clean-room, risk-first MT4 design with sensible engineering around a signal that does not have a demonstrated edge. On 16,496 EURUSD five-minute bars, the primary configuration lost 4.67%, posted profit factor 0.435, and averaged −0.389R per trade after modeled spread, commission and slippage. Every predeclared M5 variation lost money.
The failure is useful because the safety layer did what it could: fixed-fractional 0.25% risk, one position at a time, spread and volatility limits, session restrictions, a daily trend regime, daily loss and drawdown locks, and no martingale or grid recovery. Those controls can limit damage; they cannot manufacture expectancy.
Primary result
| Metric | Result |
|---|---|
| Instrument | EURUSD |
| Five-minute bars | 16,496 |
| Trades | 49 |
| Net return | −4.67% |
| Profit factor | 0.435 |
| Win rate | 46.9% |
| Maximum drawdown | 4.84% |
| Expectancy | −0.389R |
| Modeled round-trip cost | 2.1 pips |
The test used closed-bar signals, next-bar entries and a conservative rule that assigns the stop first when both stop and target appear inside the same bar. The 2.1-pip round trip represents a 1.0-pip spread, 0.7-pip commission equivalent and 0.4 pips of total slippage.
Chronological segments
| Segment | Trades | Net | Profit factor | Expectancy |
|---|---|---|---|---|
| Development | 36 | −3.66% | 0.415 | −0.413R |
| Validation | 2 | −0.24% | 0.352 | −0.497R |
| Untouched test | 11 | −0.76% | 0.530 | −0.289R |
The validation sample is too small for a precise estimate, but the conclusion does not depend on it: every segment is negative and the full-sample loss is large relative to 0.25% position risk.
Predeclared variants
| Variant | Trades | Return | PF | Max drawdown |
|---|---|---|---|---|
| Baseline | 49 | −4.67% | 0.435 | 4.84% |
| Higher reward, 1.5R | 48 | −3.36% | 0.590 | 4.12% |
| Stronger ADX filter | 31 | −3.28% | 0.400 | 3.28% |
| Deeper RSI pullback | 21 | −2.71% | 0.344 | 2.71% |
| No daily filter control | 118 | −13.20% | 0.348 | 13.37% |
The daily filter materially reduces damage, so the macro-regime idea has risk-control value. It still does not turn the entry signal positive. Removing it makes both return and drawdown far worse.
Hourly replacement screen
Rather than optimize the M5 loss surface, four hourly candidates were screened on 11,956 EURUSD H1 bars from August 2024 through July 2026. All used H4 and daily regime alignment and the same 2.1-pip cost assumption.
| Candidate | Trades | Overall return | PF | Untouched test |
|---|---|---|---|---|
| H1 pullback | 84 | −6.59% | 0.536 | −0.11% |
| H1 20-bar breakout | 85 | −1.85% | 0.855 | −2.26% |
| H1 40-bar breakout | 52 | −0.18% | 0.977 | −0.84% |
| H1 20-bar breakout, ADX 22 | 68 | −2.38% | 0.772 | −2.22% |
The 40-bar breakout is close to break-even overall, but it generated only four untouched-test trades and all four lost. Its positive validation interval therefore does not justify promotion.
Verdict
REJECTED. Do not deploy. The macro gate and circuit breakers are reasonable controls, but the pullback signal loses after modeled costs and every predeclared neighboring configuration also loses. Raising risk to pursue a monthly return target would amplify negative expectancy. No martingale, leverage increase or post-hoc parameter search is justified.
This is a research screen rather than a complete 11-gate validation: no placebo test was run, Yahoo bars are indicative mid-market data rather than executable broker ticks, and historical event-risk flags were not reconstructed. Those limitations can prevent a positive verdict; they do not rescue a strongly negative one.
Frequently asked
Did the LowDD Macro Pullback EA make money?
No. The primary EURUSD M5 screen lost 4.67% on a $10,000 starting account at 0.25% risk per trade. Profit factor was 0.435 and expectancy was −0.389R after a modeled 2.1-pip round-trip cost.
Did a higher-timeframe or higher-reward version fix it?
No. A 1.5R M5 target reduced the loss but still returned −3.36%. Hourly pullback and breakout alternatives were also negative overall; the closest, a 40-bar hourly breakout, lost 0.18% overall and failed the untouched test by 0.84%.
Is this a complete MT4 broker-tick validation?
No. It is an independent bar-by-bar research screen using Yahoo Finance indicative EURUSD bars, modeled costs and conservative same-bar execution. A final broker-specific MT4 test would still be needed, but the strongly negative screen is already enough to reject deployment rather than spend risk budget trying to rescue it.
Methodology: Independent research screen — the full 11-gate battery was not run; data and execution limits are stated in the report. Full reproducible report: backtests/lowdd-macro-pullback/REPORT.md in the source repository.Author: Validated Research Team (Methodology v1.0 — 11-gate validation). Backtests are not investment advice.