Pipsgrowth EX10005 Momentum-Scalper

Pipsgrowth EX10005 Momentum-Scalper trades fractal breakouts on XAUUSD. The signal it cares about is a structural one: the close of the most recent completed bar either breaking above the last confirmed up-fractal or breaking below the last confirmed down-fractal, in the direction that price already has relative to a 50-period EMA. Everything else in the EA exists to either qualify that breakout or survive its consequences.

How the fractal signal forms

The source defines a confirmed fractal as the kind Bill Williams originally charted — a bar whose high is the highest of the surrounding 5-bar window (up-fractal) or whose low is the lowest (down-fractal). The MT5 iFractals indicator populates those levels with a two-bar delay, which is why the EA reads from shift 2 inward, never from the forming bar. The EvaluateSignal() function copies both the upper and lower fractal buffers for DEF_FRACTAL_BARS = 5 closed bars, scans for the most recent non-empty value, and on bar 1 checks three things: is close > EMA(50) (or < for shorts), did close actually cross the last fractal level, and is the bar's body — MathAbs(close1 - open1) — at least 0.25 × ATR(14). All three have to clear before the EA reports a direction. A confidence score is computed from the body-to-ATR ratio and capped at 100, but it is only used for logging; the actual gating is binary.

Why fractal breakouts on gold

Gold is a market that trends in bursts and respects the geometry of previous swing points. Fractal breakouts capture the moment a swing high or swing low gives way, which is the moment where stops cluster and momentum tends to extend. The EMA(50) on the working timeframe provides the directional bias so the EA only buys fractal-ups and only sells fractal-downs. The 50-period EMA on H1 (DEF_HTF_TF = PERIOD_H1, DEF_HTF_EMA_PERIOD = 50) is checked by ConfirmOK(): the close of the last H1 bar has to be on the same side of the H1 EMA as the working-timeframe direction. A 20-bar tick-volume ratio of at least 0.8 (DEF_VOL_MIN_RATIO) further disqualifies the trade if the breakout is happening on a thin tape.

Regime gates

The RegimeOK() function runs before the entry is dispatched. ADX(14) main line has to be at least 18 (InpMinADX = 18.0) — below that, the market is chop and breakouts fail. ATR-percentile gating uses a 50-bar lookback (DEF_ATR_PCT_LOOKBACK = 50): the current ATR must be between the 15th and 95th percentiles of that distribution. Below 15 means dead volatility where the breakout will not pay for spread; above 95 means a news spike where the spread will widen and stops will slip. Bollinger Bands(20, 2.0) are sampled for their width; a current width below the 10th percentile of the lookback distribution is a "squeeze" — a low-conviction state the EA refuses. These three gates running together explain why EX10005 fires only on real, sustained directional moves and stays quiet through Asian-session churn.

Stop, target, and position sizing

The stop is ATR(14) × 2.0 (InpATRMultiplier = 2.0) away from entry, with a floor at the broker's stops-level. The take-profit is the same distance × InpMinRR = 1.5, so each trade is entered with a 1:1.5 reward-to-risk. Position sizing is risk-percent based: InpRiskPercent = 0.5% of effective capital, divided by the per-lot loss at the ATR stop, normalized to the symbol's lot step. The InpFixedLot = 0.01 is the fallback when risk is set to zero or the lot computation hits a degenerate case. OrderCalcMargin pre-checks that the resulting order fits inside free margin before any trade.Buy() or trade.Sell() call goes out.

Position management in flight

ManageTrades() runs on every tick for every open position. It checks for opposite-signal exits first — if the EA sees a fractal-down signal on bar 1 while holding a long, it closes immediately. It also closes if the ADX(14) collapses below 18 (logged as "regime-change-ADX"). After those, a time exit fires at DEF_MAX_BARS_IN_TRADE = 240 bars (roughly 20 hours on M5). At 1R of profit, the stop is moved to break-even plus 5 points (DEF_BE_OFFSET_POINTS = 5). At 1R, 50% of the position is closed in a one-shot partial (DEF_PARTIAL_FRAC = 0.5). At 1R, an ATR(14) trail at 2.5× ATR begins, ratcheting the stop behind price without ever moving it backward. The partial TP is marked in the deal comment with [P] so it never re-triggers.

Pyramid mode (off by default)

InpEnablePyramid = false is the shipped default. When toggled on, the EA looks for its most profitable open position of either type, refuses to act unless that position is in profit, and adds a new entry only if the current price has moved at least DEF_PYRAMID_SPACING_ATR = 1.5 ATRs beyond the original open. Hard cap is DEF_MAX_PYRAMID = 3 per direction. RegimeOK() is rechecked before each add. There is no martingale and no lot escalation — the new leg is sized by the same risk formula as the first.

Loss-cap and cooldown architecture

UpdateRealizedPnL() rebuilds the daily and weekly realized P&L from HistoryDealsTotal() on every tick, filtering by magic and symbol, summing profit plus swap plus commission. Daily cap is 3% of effective capital (InpDailyLossLimitPercent = 3.0); weekly cap is 6% (InpWeeklyLossLimitPercent = 6.0). If either trips, no new entries are sent for the rest of the day or week. The cooldown layer watches the last hour of deal history; three consecutive losses (DEF_COOLDOWN_LOSSES = 3) push the EA into a 12-bar no-entry state (DEF_COOLDOWN_BARS = 12). A separate capital floor at InpCapitalCapFloor = 50.0 USD halts entries when effective equity drops below it.

Order-execution safety

OnInit() reads SYMBOL_FILLING_MODE and picks FOK, IOC, or RETURN in that order. trade.SetDeviationInPoints(20) caps the slippage the broker is allowed to charge. The spread filter is dynamic: the default ceiling is 250 points (DEF_MAX_SPREAD_POINTS = 250), but the EA also reads SYMBOL_TRADE_STOPS_LEVEL and, if the broker's stops-level is below 250, the spread cap is reduced to 4× the stops-level so a tight-stop entry can never be sent into a wider spread than its own protective distance. InpDryRun = true is the shipped default; with it on, the EA logs what it would have sent but does not place orders — the right starting posture for any broker-test or prop-firm trial.

When it should run and where

The session filter is server-time 7 to 21, the windows where London and New York overlap and XAUUSD is in genuine directional flow. Hours 12 to 14 are a soft news blackout (DEF_NEWS_START_HOUR = 12, DEF_NEWS_END_HOUR = 14) — non-feed, time-based only, so it will quiet the EA across the regular US-data release slot. Weekends are blocked. The OnTester formula is (net × profit-factor) / (1 + relative-DD) with a 30-trade minimum — that is what the MT5 Strategy Tester's optimization graph will rank the inputs by, so any parameter sweep is being scored on net × pf / (1 + DD) and not raw profit.

Bottom line

EX10005 is a regime-gated, fractal-breakout scalper for XAUUSD. It is conservative about when it trades (ADX/ATR-percentile/BB-width gates all have to clear), aggressive about how it manages a trade once on (1R BE, 50% partial, ATR trail), and quiet the rest of the time. With pyramid off, max three positions, and a 240-bar time exit, the worst-case live profile is well-bounded; the only requirement is a broker that keeps gold's spread inside the 250-point cap and the stops-level tight enough that the spread filter does not choke the entries.