Pipsgrowth EX18028 TrendFollow

TrendRider XAUUSD 5M is built around a deliberate two-vote signal: a Heikin-Ashi closed-bar color vote and an inline Hull moving-average slope vote. Both votes are computed on the most recently closed M5 bar (index 1), and the EA only fires when they agree, when the regime engine tags the market as tradeable, and when the higher-timeframe EMA pair confirms direction. The HA reconstruction is done in EA — CalcHAAt() walks bar-by-bar from a seed ((open+close)/2 for the prior HA_Open, OHLC4 for HA_Close) and propagates (prev_open+prev_close)/2 forward — so there is no dependency on the Heikin-Ashi indicator. The Hull is also hand-rolled: CalcHullSeries() reads two iMA(..., MODE_LWMA) buffers (half period = InpHullPeriod/2 = 25, full = 50), takes 2 * WMA_half - WMA_full, then smooths that difference with a final ManualLWMA() over sqrt(50) ≈ 7 bars. The resulting hull series is queried at index 1 (current closed bar) and index 2 (prior closed bar) — a positive slope (hull1 > hull2) is the bullish vote, a negative slope the bearish vote.

The signal confidence is graded. When both HA and Hull vote the same direction (both +1 or both -1), GetClosedBarSignal() returns signal = ±1, confidence = 80. When only HA has a clean color and the Hull is flat (no slope), confidence drops to 50 — the entry is taken but with a lighter conviction flag. When the two votes disagree the function returns signal = 0 and no trade is considered; this is the primary guard against HA-flip chop with a lagging Hull. The vote result is then handed to ConsiderEntry() which runs the full filter stack before any order is sent.

The regime engine is a six-state classifier that runs on every tick. UpdateRegime() reads ADX(14), ATR(14), and Bollinger Bands(20, 2.0) and computes two rolling percentiles over a 100-bar lookback: the ATR percentile (g_atr_pctl) and the Bollinger-width percentile (bw_pctl). The classification is: ADX >= 40 → REGIME_STRONG_TREND; ADX >= 25 → REGIME_WEAK_TREND; bw_pctl >= 80 or g_atr_pctl >= 75 → REGIME_EXPANDING; bw_pctl <= 20 → REGIME_COMPRESSING; g_atr_pctl <= 25 → REGIME_CHOPPY; otherwise REGIME_RANGE. Only three of these six — STRONG_TREND, WEAK_TREND, EXPANDING — are tradeable per RegimeTradeable(). CHOPPY, RANGE, and COMPRESSING all block new entries, and a flip into any non-tradeable state force-closes open positions via the (j) regime-change exit branch in ManageOpenPositions().

The higher-timeframe confirmation is an EMA(20)/EMA(50) pair on the timeframe set by InpHTFTF (the documented 1–7 range maps to M1..D1; an out-of-range value — including the default 8 — falls through to PERIOD_H1 in MapTimeframeInt()). HTFTrendAgrees() returns true when the HTF fast EMA is on the correct side of the slow EMA relative to the signal. A long signal needs EMA_20_H1 > EMA_50_H1; a short needs the opposite. This filter is what makes the EA's M5 entries ride H1 trends rather than fade them.

Position risk is sized off effective capital — the minimum of account equity and the InpCapitalCapAmount capital-allocation cap (set to 0 by default, which disables the cap and uses the full equity). A 0.5% per-trade risk (InpRiskPercent = 0.5) is translated into lots via CalcLotByRisk(), which uses the symbol's tick value and tick size. A $50 floor (InpCapitalCapFloor) blocks trading when effective capital drops below it, so a small account cannot be ground down by repeated micro-lot losses. The lot is then normalized to the symbol's lot step and capped by LotsMin/LotsMax, and a margin pre-check via OrderCalcMargin confirms free margin covers the order.

The exit model is multi-stage. On entry, a single stop is set at 2.5 × ATR(14) below the ask (longs) or above the bid (shorts) — InpATRMult = 2.5 is also the ATR trailing-stop multiple, so the initial risk distance and the trail distance are identical. The take-profit is 1.5 × risk_distance (the InpMinRR = 1.5 minimum reward:risk filter), giving a 1:1.5 R:R structure. The broker's stops-level is used as a floor if the ATR-derived distance is too small. While the trade is open, ManageOpenPositions() runs on every tick and applies: (a) a one-shot break-even at 1R, raising the stop to entry plus a five-point buffer; (b) a 50% partial close at 1.5R (tracked per ticket via g_partial_done[] so it fires exactly once); (c) an ATR-trailing ratchet that only moves the stop forward, never backwards; (d) a momentum-decay exit where RSI(14) on the closed bar drops below 45 (longs) or rises above 55 (shorts); (e) an opposite-signal exit when the closed-bar vote flips; (f) a 60-bar time exit (5 hours on M5) via InpMaxBarsInTrade = 60; and (g) the regime-change force-close. The seven exit conditions are checked in this fixed order each tick.

Risk guards beyond position sizing include a 3% daily loss limit on effective capital (the EA scans the deal history and trips g_trading_disabled for the rest of the day once realized PnL hits -3%); a cooldown that arms after 3 consecutive losing trades and blocks new entries for 30 minutes (COOLDOWN_LOSSES = 3, COOLDOWN_MIN = 30); a 50-point spread cap (InpMaxSpreadPoints); a configurable session window (InpSession = "08:00-20:00" server time) that blocks entries outside the window; a 120-minute pre-Friday-close flatten (InpFridayCloseMinutes = 120); a 5-minute news blackout around high-impact events pulled from the MQL economic calendar via CalendarValueHistory(); a duplicate-bar guard so the same closed bar cannot trigger two entries; and InpKillSwitch to flatten and stop. InpDryRun defaults to true — the EA logs every would-be trade without sending orders until you flip it off.

The OnTester() function customizes the optimization criterion: (net * profit_factor) / (1 + balance_dd) with a 10-deal floor, which means the strategy tester will reject parameter sets with fewer than 10 closed trades during optimization. The default pyramid setting (InpPyramidMaxLevels = 0) hard-caps scaling to OFF, so this EA is a strict single-position system per magic on the symbol — concurrent trades in the opposite direction are explicitly blocked.

For live deployment: this EA is built for XAUUSD on M5 with a $100 minimum deposit and a MEDIUM risk profile. Use a broker that quotes tight gold spreads (the 50-point filter is generous, but a 15–25 point spread gives the entries headroom) and a server whose clock matches the configured InpSession window — many brokers run GMT+2 or GMT+3 server time, and the 08:00–20:00 default captures the full London–New York overlap. The EA has no martingale, no grid, and no averaging-in. The Hull(50) period is large for M5 and produces a deliberate lag that the regime filter is designed to compensate for; if you drop the period much below 30, expect more whipsaws and lower confidence votes.