Pipsgrowth EX10006 Momentum-Scalper

EX10006 reads the M1 USDJPY tape through a stack of three exponential moving averages and trades only the micro-pullbacks that occur inside an already-established trend. The slope filter is the 9-period vs 30-period EMA cross on the working timeframe, with the 5-period EMA carving out the pullback zone. A long setup requires the 9 above the 30, the previous bar's low coming within 1.0× ATR(14) of the 5-EMA, the bar closing up versus the bar before (a one-bar momentum candle), and the bar's low clearing the lowest low of the trailing 12 bars — a higher-low structural bias. The short side is the mirror image: 9 below 30, bar-1 high within 1.0× ATR of the 5-EMA, bar-1 closing down versus bar-2, and bar-1 high below the highest high of the trailing 12 bars. The pattern is buy-the-dip-in-an-uptrend, not momentum chase, and it deliberately fires on closed bars only to avoid live-bar repaint.

The signal layer produces a confidence score from 0 to 100, derived from the bar-1 versus bar-2 close delta normalized by ATR(14), but the score is logged for audit purposes only and does not gate the entry. The only candle-shape requirement is the bar-1 vs bar-2 close direction; there is no engulfing, no pin, no inside-bar, no wick-ratio, no pattern detector in the stack. This is by design — the entry is a geometry of moving averages plus a single momentum candle, nothing more.

Before any order reaches the market, three pre-trade gates run in sequence. RegimeOK() reads ADX(14) on the working timeframe and rejects entries when ADX is below the configurable threshold InpMinADX (15.0 default), when the current ATR-percentile over a 50-bar lookback is below 15 (dead volatility) or above 95 (spike volatility), or when the Bollinger Band width-percentile is below 10 (a dead squeeze). ConfirmOK() reads a 50-period EMA on M15 and requires the previous M15 close to be on the correct side of the 50-EMA for the signal direction; the same gate reads the M1 bar-1 tick volume and rejects entries below 0.8× the rolling 20-bar volume average. The third gate, embedded in TryEntry(), enforces the position-count cap (default InpMaxConcurrent = 2 plus a 3-position symbol-exposure hard cap from DEF_MAX_SYMBOL_EXPOSURE), the one-per-direction cap so a buy and a sell cannot coexist on the same symbol, the capital floor (InpCapitalCapFloor = 50 USD by default, blocks new entries only), the 3% daily and 6% weekly loss caps computed against the effective capital, the active cooldown window after 4 consecutive losses, and the standard broker trade-permission flags. OrderCalcMargin pre-checks the lot against free margin before any send.

Position sizing is computed in CalcLotForATRStop(). The current ATR(14) is multiplied by InpATRSLMult (1.2 default) to get the stop distance and by InpATRTPMult (0.6 default) to get the target distance. The result is the EA's signature pattern: a 0.5 reward-to-risk scalp — 0.6R target against a 1.2R stop, deliberately sub-1R so the high hit-rate carries the equity curve. Lot size comes from InpRiskPercent (0.5% of effective capital default) divided by the loss-per-lot at the ATR stop, floored to the symbol's lot step, and clamped to the min and max lot bounds. The InpFixedLot 0.01 fallback engages when risk-percent is zero or the loss-per-lot calculation returns non-positive.

Initial stop-loss is placed at entry ± 1.2× ATR(14) with a hard floor at the broker's minimum stop-level distance (read from SYMBOL_TRADE_STOPS_LEVEL and multiplied by point). Initial take-profit is placed at entry ± 0.6× ATR(14) with the same floor. Once the position is open, ManageTrades() runs every tick and applies a four-stage ladder. At 1.0R of profit, the stop ratchets to break-even plus 5 points — a one-shot adjustment that re-arms only on a new entry. At the same 1.0R threshold, a partial close takes 50% of the position volume, and the comment is annotated with [P] so the partial never re-fires on the same ticket. Beyond 1.0R, the trail activates at 2.5× ATR(14) and only tightens; it never loosens. The fourth stage is the time-stop: any position that has been open for DEF_MAX_BARS_IN_TRADE (60 bars, one hour on M1) is closed regardless of price.

Two independent exit paths run in addition to the SL/TP. If EvaluateSignal() flips direction while a position is open, the position is closed with an opp-signal tag. If ADX collapses below 15 during the trade, the position is closed with a regime-change-ADX tag. Both paths exist to cut trades whose underlying thesis has broken, even before the stop-loss is hit. The intent is to avoid giving back unrealized gains in a trade whose signal logic has already invalidated itself.

The session filter is straightforward. The EA trades only when the server hour is between 0 and 23 inclusive (effectively 24-hour coverage, gated to weekdays by a Sunday/Saturday check), skips Friday hours at or after DEF_FRIDAY_CUTOFF_HOUR (21 default), and applies a soft 12:00–14:00 news-blackout window via DEF_NEWS_START_HOUR and DEF_NEWS_END_HOUR. The news-blackout is a time-window filter — it does not pull from the MQL5 economic calendar feed, so high-impact releases outside the 12:00–14:00 server window are not blocked. A typical USDJPY M1 trader running this EA will see real activity from Asian open through New York close, with a hard halt before the weekend gap.

The pyramid module is fully implemented in TryPyramid() but shipped off by default (InpEnablePyramid = false). Setting it to true enables the path: it only adds to a position already in profit, only when the current price has moved at least DEF_PYRAMID_SPACING_ATR (1.5 default) from the original open, and only when the regime gate still passes. The hard cap is 3 levels per direction (DEF_MAX_PYRAMID). The pyramid leg is sized through the same CalcLotForATRStop() function as the initial entry, so the SL/TP geometry stays consistent. The intent is the standard scale-into-a-winner pattern, but the default is conservative because compounding from a bad first entry is a real risk on a 0.5R system.

The risk layer has two additional safety mechanisms beyond the daily/weekly caps. InpCapitalCapAmount (0 default, i.e. disabled) sets a hard dollar ceiling on effective capital — used to ring-fence the EA's allocation when running on a multi-strategy account. InpCapitalCapFloor (50 default) blocks any new entry when effective capital drops below that line; the EA continues managing open positions but does not open new ones. The kill switch (InpKillSwitch = true) closes every position this EA owns on the next tick — a manual panic button. Dry-run mode (InpDryRun = true default) prevents all live sends and prints the intended order to the journal; the EA ships in dry-run by default so the strategy can be audited before any capital is exposed.

The OnTester custom criterion is the family-standard formula: net * profitFactor / (1.0 + balance_dd_percent), with a 30-trade minimum and a guard that returns 1.0 for non-positive pf or dd. This favors strategies with positive net profit, high profit factor, and low relative drawdown — exactly the metrics a 0.5R scalper should optimize for in the strategy tester. A 0.5R system with 60% hit-rate will post a much higher custom criterion than a 1.5R trend-follower with 40% hit-rate at the same net profit.

Order execution uses an auto-detected fill mode: the broker's SYMBOL_FILLING_MODE mask is read and tried in FOK → IOC → RETURN order. Slippage is capped at 20 points via DEF_DEVIATION_POINTS. The initial send either fills or the journal logs an [EXEC-FAIL] with the retcode and description — there is no outer retry loop on the entry send. On REQUOTE, PRICE_OFF, or TIMEOUT retcodes, the EA refreshes rates and resends once. Close operations route through TryClose_EX10006, which retries up to 3 times on REQUOTE / TIMEOUT / PRICE_OFF / PRICE_CHANGED with a 200 ms backoff. Partial closes route through TryClosePartial_EX10006 with the same retry policy. Modify operations route through TryModify_EX10006, which retries up to 3 times on REQUOTE / TIMEOUT with a 100 ms backoff.

In backtest on USDJPY M1 with default inputs, expect a high-frequency equity curve with many small wins and occasional larger drawdowns from the 0.5R negative tail. The 12-bar structural bias filter should keep most losses contained to the 1.2R stop distance. Pyramiding off by default means a single bad signal cannot snowball — every position is a discrete bet. The 60-bar time-stop means even a poorly-timed entry that drifts will be closed before it can become an overnight hold.