Pipsgrowth EX10007 Momentum-Scalper

Pipsgrowth EX10007 Momentum-Scalper is a closed-bar trend-pullback scalper built around an EMA(50) over EMA(200) macro trend, with RSI(14) crossing the 50 midline as the entry trigger and a hard regime gate in front of every order. The header advertises twelve cooperating layers, but the implementation actually exercises seven of them — REGIME, SIGNAL, CONFIRM, ENTRY, SIZING, MANAGE/EXIT, plus the OnTester fitness function. CAP, RISK, NO-TRADE, and SCALING are present as guard checks rather than as separate modules. Reading the .mq5 directly shows where the layers start and stop, and that gap between marketing and implementation is worth understanding before you trust the equity curve.

The signal engine is the heart of the EA. EvaluateSignal() reads the last two closed bars of iMA(_Symbol,_Period,50,0,MODE_EMA,PRICE_CLOSE), iMA(_Symbol,_Period,200,0,MODE_EMA,PRICE_CLOSE), and iRSI(_Symbol,_Period,14,PRICE_CLOSE). A long needs all six conditions simultaneously: the fast EMA is above the slow EMA, RSI crosses from below 50 to above 50 between the previous and last closed bar, the close of the last closed bar sits above the fast EMA (a pullback reclaim), the fast-EMA slope in points is at least DEF_MIN_EMA_SLOPE_POINTS = 10.0, the absolute distance between the two EMAs is at least DEF_MIN_EMA_DIST_ATR = 0.5 of the current ATR(14), and the close exceeds the previous bar's high (the breakout confirmation, governed by InpUseBreakoutConfirm = true). The short side mirrors that stack. None of these conditions is discretionary — confidence is then computed as min(100, (slope / 20) * 50 + 25) and forwarded to TryEntry().

Before any signal is even considered, RegimeOK() requires a real trend to be present. The function pulls iADX(_Symbol,_Period,14) and rejects the bar if ADX main line is below InpMinADX = 20.0. It then computes the ATR-percentile against a 50-bar lookback: bars with current ATR below 15% of the recent average are rejected as dead volatility, and bars above 95% are rejected as spike events. Finally, it inspects the iBands(_Symbol,_Period,20,0,2.0,PRICE_CLOSE) width and refuses entries during a BB-width squeeze (current width below 10% of the 10-bar average). This three-stage regime check is what kills the bulk of the false breakouts that pure EMA-cross systems suffer from — the EA simply does not try to trade when the market is compressing or when directional energy has leaked out.

ConfirmOK() adds two more gates. When InpHTFConfirm = true is on, the EA loads a second pair of EMAs on PERIOD_M15 (DEF_HTF_TF) and requires the M15 fast-EMA/slow-EMA relationship to agree with the M5 signal. This is the most consequential toggle in the inputs: with HTF confirm off, the EA will fire M5 signals even when the M15 trend is the other way, and you can expect a noticeably larger trade count at the cost of a noticeably lower win rate. The second confirmation is a tick-volume filter — current tick volume must be at least 80% of the 20-bar rolling average (DEF_VOL_MIN_RATIO = 0.8). The volume check uses raw iVolume() of bar 1; it will be near-zero on FX symbols that don't report tick volume, in which case the comparison is skipped and the filter is effectively neutral.

The trade execution path is honest about partial-fill risk. OnInit() reads SYMBOL_FILLING_MODE and assigns FOK → IOC → RETURN in that order, so the EA adapts to whatever your broker supports. Slippage is set to 40 points via DEF_DEVIATION_POINTS (40 points on XAUUSD is roughly $0.40, or 4 pips). After the first trade.Buy/trade.Sell, if the result retcode is TRADE_RETCODE_REQUOTE, TRADE_RETCODE_PRICE_OFF, or TRADE_RETCODE_TIMEOUT, the EA refreshes the symbol quote and re-sends once. Closing and modification go through dedicated TryClose_EX10007, TryClosePartial_EX10007, and TryModify_EX10007 helpers, each of which loops up to three times with a 200ms sleep on retcodes 10004/10006/10007/10009. There is no walk-the-book logic — when the third retry fails, the action is logged and abandoned.

Stop placement is fully ATR-driven. CalcLotForATRStop() measures the stop distance as InpATRMultiplier * ATR(14) with a default of 1.5× ATR, then floors it at the broker's SYMBOL_TRADE_STOPS_LEVEL. The take-profit is set at stopDist * DEF_TP_ATR_MULT = 1.3, which yields a fixed TP-to-SL ratio of 1.3 to 1.5 (R-multiple 0.87) on the initial order. The asymmetric sizing is deliberate — the EA expects to make most of its money on the trail, not on the initial target. When InpRiskPercent = 0.7 is enabled, the position size is computed as riskMoney / lossPerLot, where lossPerLot = (stopDist / tickSize) * tickValue, so a $1,000 account with a 0.7% risk on a 1.5× ATR(14) stop lands in the 0.02–0.05 lot range on most XAUUSD brokers. Setting InpRiskPercent to 0 or negative forces the EA to use InpFixedLot = 0.01 flat.

Once a position is open, ManageTrades() runs on every tick and stacks three independent exit conditions. When unrealized profit reaches 1.0R (DEF_BE_TRIGGER_R = 1.0), the stop is moved to entry + 15 points offset, locking in a scratch-or-better outcome. At the same profit threshold, if the position is more than 1.5× the minimum lot, half the position is closed via ClosePartial() and tagged [P] in the comment to prevent re-triggering. After 1.0R profit, the trailing stop activates at 1.0× ATR behind price (DEF_TRAIL_ATR_MULT = 1.0). Beyond management, two structural exits run first: a closed-bar EvaluateSignal() call that closes longs on a fresh sell signal and shorts on a fresh buy signal, and a time-based exit at DEF_MAX_BARS_IN_TRADE = 240 bars. The 240-bar cap is 20 hours of M5 bars, so an M5 trade that survives 20 hours without trailing out gets force-closed regardless of P&L. Regime collapse (RegimeOK(0) returning false on ADX) also closes the position with a regime-change-ADX reason tag.

The risk and safety stack is conservative even though the EA is labeled HIGH. The 0.7% per-trade risk is light, and the loss-limit circuit breakers are tight: 5% of effective capital per day and 10% per week stops the EA before catastrophic damage. The capital cap is opt-in: with InpCapitalCapAmount = 0 (the default), no cap is applied and the EA sizes off full equity; setting it to e.g. 5000 means the EA treats the first $5,000 of equity as its trading capital and any excess as out-of-bounds. A floor check (InpCapitalCapFloor = 50.0) blocks new entries when effective capital falls below $50. A cooldown of 3 bars (15 minutes on M5) engages after 3 consecutive losses — UpdateConsecutiveAfterClose() reads the recent deal history to detect losing exits, increments a counter, and arms g_cooldownUntil. Note one operational caveat: the kill switch (InpKillSwitch = true) is a panic-close-all-on-tick — flipping it live will close every position the EA owns, so use it from the inputs panel, not by code change.

Two non-obvious behaviors are worth knowing about. First, InpDryRun = true is the default — the EA will print [DRY-RUN] entries to the journal but not send any order. You must explicitly set InpDryRun = false before going live. Second, the spread filter uses DEF_MAX_SPREAD_POINTS = 60 (6 pips on XAUUSD) but then self-corrects: if the broker's SYMBOL_TRADE_STOPS_LEVEL is lower, the cap is widened to 4× the stops level; if the stops level is higher, that becomes the effective cap. A secondary ratio filter refuses entries when ATR(points) / spread < 3.5, which on a 60-point XAUUSD spread ceiling means the broker must deliver at least 17-point ATR — non-issue on gold, but worth knowing if you adapt the EA to lower-volatility pairs.

The OnTester fitness function is the standard (net * profitFactor) / (1 + balanceDD%), returning 0 when there are fewer than 30 trades so a single lucky entry can't dominate the optimization pass. With pyramid disabled (the default), expect the strategy tester to settle into 1–3 trades per day on XAUUSD M5 during London/NY, with a win rate in the 40–55% band but a positive expectancy because the partials + trail combination converts small winners into runners. Backtests on a cent account with $100 will move the needle; backtests on a $10,000 standard account will need a longer data window before the per-trade dollar noise averages out.