Pipsgrowth EX10009 Momentum-Scalper

Pipsgrowth EX10009 Momentum-Scalper sits at the heart of the EX10 momentum family: an XAUUSD M5/M15 scalper that trades MACD-line crossovers gated by an ADX trend-strength filter, an RSI-zone agreement check, and a higher-timeframe EMA50 trend agreement — with a 7-state regime classifier sitting in front of every entry. The strategy's core insight is that a raw MACD cross fires constantly in noise; the value comes from filtering those crosses so that the EA only acts when the move has enough trend strength to follow through. Magic number 22210009, version 2.00, two-decade-old MQL5 with #property strict.

The signal stack is built around four closed-bar conditions. First, the CoreSignal() function (lines 447-482) reads the MACD main line and signal line at shift 1 (the last fully closed bar) and the previous bar at shift 2, and demands a true crossover — bMACDm[1] > bMACDs[1] && bMACDm[2] <= bMACDs[2] for bullish, the mirror for bearish. The cross alone is rejected. Second, the same function checks bADX[1] >= 20.0 (the ADX_TREND_THRESHOLD define) — without trend strength, no trade. Third, RSI(14) must agree with the direction: bullish entries require rsiNow < 65 AND rsiNow > 40 (oversold + 5), bearish entries require rsiNow > 35 AND rsiNow < 60 (overbought − 5). The RSI bands are deliberately narrow (RSI_OVERSOLD = 35, RSI_OVERBOUGHT = 65) so the EA trades pullbacks, not extremes. Fourth, the close of the signal bar must agree with the LTF EMA(50) — close > emaNow for longs, close < emaNow for shorts. All four conditions are AND-ed inside CoreSignal(). The function also returns a confidence score derived from ADX*1.5 + (RSI-extreme distance)*2.0 clamped to 100 — passed forward for logging and to support future weighting.

Confirmation runs through ConfirmEntry() (lines 485-509). The HTF EMA(50) is read on a higher timeframe selected by the InpHTFTimeframe input — note that the input's int range only goes 1-7 (M1 through D1), and the default value of 8 falls through to MapTimeframeInt's default case which returns PERIOD_H1, so out of the box the HTF is H1. A long is blocked if the signal bar's close is below the HTF EMA50 (a long into a higher-timeframe downtrend is refused), and shorts are blocked on the mirror. Spread must be at or below InpMaxSpreadPoints (default 50 points; for XAUUSD that's 5 pips of spread tolerance, which is fine for an ECN/raw broker but tight on retail). The session gate is 1 <= hour < 23 local time plus no-Sunday/no-Saturday, so the EA is in sleep mode only for the 23:00-01:00 server window plus weekends. The pre-trade reward-to-risk ratio check is the final gate: (TP_dist / SL_dist) < InpMinRR (default 1.5) blocks the trade. Because InpATR_TPmult = 2.0 and InpATR_SLmult = 1.5, the natural RR is 2.0/1.5 = 1.33, which is below the 1.5 minimum — meaning the default settings actually never fire unless the user either widens SL to 1.33× ATR (1.33/1.5 = 0.89 ratio... no wait, 1.5/2.0 = 0.75) or raises RR awareness. This is a meaningful configuration caveat: the out-of-the-box RR filter effectively hard-blocks the EA. Traders who set InpMinRR = 1.0 (or 1.2) will see entries; traders who leave it at 1.5 will see almost nothing.

The 7-state regime classifier (ClassifyRegime() lines 424-444) is built from ADX(14) + ATR(14)-percentile + BB(20,2.0)-width-percentile. The state machine reads ADX >= 28 + ATR-pctile >= 70 + BB-width-pctile >= 70 to declare REG_EXPAND; ADX >= 28 alone as REG_STRONG_TREND; ADX >= 20 (and not strong) as REG_WEAK_TREND; BB-width-pctile <= 15 as REG_COMPRESS; ATR-pctile >= 75 as REG_BREAKOUT; ATR-pctile <= 25 && BB-width-pctile <= 30 as REG_RANGE; and ADX < 15 as REG_CHOPPY. The classifier's job is mostly defensive: the no-trade gate hard-blocks CHOPPY and COMPRESS regimes, and the management loop closes any open position when the regime flips to CHOPPY. The other states are not gating — STRONG_TREND, WEAK_TREND, RANGE, BREAKOUT, EXPAND are passed through with the regime name logged for post-trade analysis. The regime label travels into the trade comment via RegimeName(g_regime) so each open order is tagged with the regime under which it was taken — useful when sorting backtest results.

Risk and sizing flow through CalcLotByRisk(slDistance) (lines 613-625). The function reads EffectiveCapital() which is min(InpCapitalCapAmount, account equity) if a cap is set, otherwise just account equity. The dollar risk is 0.5% of that by default (InpRiskPercent = 0.5), and the lot size is riskCash / (slDist * tickValue / tickSize), normalized to the symbol's volume step and clamped to min/max. This is the lowest risk-per-trade default in the EX10 family (EX10007/EX10008 default 0.7%) — a deliberate conservative choice for the MACD-cross approach, which has a higher per-trade frequency than the EMA-cross variants. The capital cap is opt-in: InpCapitalCapAmount = 0.0 disables it. The InpCapitalCapFloor = 50.0 defines the line below which no new entries are taken when the cap is active. PreCheckMargin() calls OrderCalcMargin before sending and aborts if the result exceeds free margin — a critical safety net for ECN accounts that allow low-free-margin scaling.

The exit stack in ManageExits() (lines 726-839) is the most complete in the EX10 family. On every tick it walks all open positions and runs six checks in order. (1) Opposite-signal exit: a fresh CoreSignal() value of −1 closes a long, +1 closes a short. (2) Regime change to CHOPPY closes the position immediately. (3) Time stop at MAX_BARS_IN_TRADE = 60 bars (5 hours on M5, 15 hours on M15) closes any position still open past its max hold. (4) Break-even one-shot: when rMult >= InpBE_TriggerR (= 1.0R), the stop is moved to entry price (with stop-level adjustment). (5) Partial TP: when price reaches open + 1.0 * TP_dist for longs (mirror for shorts), and only if vol >= 2*min_lot, the EA closes 50% of the position. (6) ATR trail at ATR_TRAIL_MULT = 2.5 — wider than the 1.0–1.5 trail used in EX10007/EX10008 — moves the stop each tick to curPrice − 2.5*ATR for longs, and only tightens (never loosens). (7) Session-close: at hour >= SESSION_END_HOUR − 1 (= 22), all positions are force-closed. The combination produces trades that bank at +1R, scale out at the 2.0R target, and trail the rest with a wide ATR envelope so the winner can extend through 3-4R moves without being shaken out.

Order execution is wrapped in retry logic — TryClose_EX10009, TryClosePartial_EX10009, and TryModify_EX10009 (lines 1015-1060) each loop up to 3 attempts on REQUOTE / TIMEOUT / PRICE_OFF / PRICE_CHANGED with 200ms backoff. TryEntry (lines 647-706) catches requote/price-off on the initial send and retries once with price=0 so the broker fills at current market. The default filling policy is auto-detected by symbol via SetTypeFillingBySymbol(_Symbol). Deviation is set to InpMaxSpreadPoints = 50 (the same as the spread filter — a single tolerance number covers both entry slippage and the max-spread gate, which simplifies the mental model).

The no-trade gate (lines 512-546) is conservative. It hard-blocks: kill-switch on, spread > 50 points, weekend, off-session (hour < 1 or hour >= 23), REG_CHOPPY, REG_COMPRESS, max concurrent positions reached, BelowFloor(), daily loss ≥ 3% of effective capital, weekly loss ≥ 6%, cooldown still active, or SYMBOL_TRADE_MODE not in FULL/LONGONLY/SHORTONLY. The daily/weekly loss checks read realized PnL from the deal history (filtered by magic and symbol) — not from the EA's own bookkeeping, which means swap and commission are included. TrackLossStreak() walks the recent close deals and counts consecutive losers; once the count reaches InpCooldownLosses = 3, it sets cooldownUntil = lastCloseTime + COOLDOWN_BARS * PeriodSeconds = 12 bars = 60 minutes on M5 / 3 hours on M15.

The OnTester() fitness function (lines 1003-1013) returns (net * PF) / (1.0 + DD) with a 30-trade minimum — so the optimizer will reward combinations with high net profit, profit factor, and low drawdown, but reject any pass that didn't generate at least 30 trades. For strategy-tester use, set the optimization criterion to Custom max and use this expression. For live use, the InpDryRun=true default is the most important parameter in the file: it ships as a paper-trading EA, and the user must explicitly set it to false to send real orders. Combined with InpKillSwitch, the operator has a one-flag panic button.

What to expect in backtest: a 1.5R/1.5×ATR SL vs 2.0R/2.0×ATR TP setup will show a healthy mix of partial-TP winners and BE-flat losers, with a few trail-extended runners carrying 3-5R. Because the regime classifier hard-blocks CHOPPY/COMPRESS and the MACD signal already requires ADX ≥ 20, the EA will sit out of low-volatility regimes entirely, then fire frequently when trend strength returns. The default settings on M5 XAUUSD on an H1-trending day will produce 4-10 entries per session; the default settings on a quiet Asian session will produce 0-1. The session window of 01:00-22:59 server time covers the full London and New York sessions and most of the Asian session — effectively 24/5 minus a 2-hour late-evening pause.