Pipsgrowth EX18051 TrendFollow

EX18051 is built around an uncommon twist on the standard Hull Moving Average: instead of computing a single Hull line from close prices, it runs three parallel Hull instances — one fed by the high of each bar (HullH), one by the low (HullL), and one by the close (HullC). Each instance calculates its own WMA of the full period, the WMA of half the period, twice the half minus the full, then a final WMA smooth of the result with period equal to the square root of the base period (the textbook Hull construction). Each Hull is asked to color itself green, pink, or flat by comparing the current bar output to the prior bar output against the threshold InpHullThresh (0.0001 price units, deliberately tiny so noise gets counted as flat). The signal function HullSignal() counts how many of the three lines are green on the most recent closed bar and how many are pink. If at least InpReqAgree (default 2) of them are green AND the close Hull has remained green across the last 3 closed bars, the EA reads a long bias. The same test in pink yields short bias. The 3-bar stability requirement is the EA's main defense against whipsaw: a single bar of disagreement breaks the streak and the signal is dropped, which means a flip from green to pink needs the close Hull to hold pink for three consecutive bars before any short trade is taken.

The signal alone is never enough. Every candidate trade has to clear ConfirmDir(), a 2-of-3 panel of confirmations. The first vote comes from an inline Heikin-Ashi transform: the EA builds HA open and close arrays from the OHLC bars (passing close for the open array as a smoothing approximation), then asks whether the most recent closed HA bar is green for longs or pink for shorts. The second vote is the inline SuperTrend: an ATR(22) channel with multiplier 3.0 wraps price, and the trend flag flips when close breaks the opposite band. The third vote is the higher-timeframe EMA(50) on the user-selected HTF (default H4 per InpHTF=6). For a long, the current HTF close must sit above the EMA(50); for a short, below. Two of these three must agree, so a pure Hull signal with no corroboration from any of HA, SuperTrend, or HTF EMA is rejected before it ever reaches the risk layer.

Regime detection runs every tick from a 5-state classifier: STRONG_TREND when ADX is at or above 25 and ATR-percentile is at least 60; WEAK_TREND when ADX is at least 20 and ATR-percentile at least 40; BREAKOUT when Bollinger bandwidth is wider than 10% of the midline and ATR-percentile is at least 70; RANGE when BB width collapses below 4%; and CHOPPY as the catch-all when nothing else fits. CHOPPY blocks new entries outright and forces the ManageExits() pass to close any open position attributed to this magic; RANGE blocks entries too. So the EA will only put on trades when the market is actually moving in a usable way.

Risk is sized off an effective capital figure that is either the live account equity or, when InpCapAmt is set above zero, the smaller of the cap amount and the equity. A floor of InpCapFloor (default $50) blocks trading entirely when the cap or equity would fall below it. Each trade risks InpRiskPct (0.5%) of that effective capital, with the stop loss distance calculated as 1.6× ATR(14) on the closed bar. Lot size comes from CalcLots(), which divides risk money by the per-lot loss at the chosen SL distance using the symbol's tick value and tick size, then normalizes to the broker's volume step, minimum, and maximum. A margin pre-check calls OrderCalcMargin() and refuses to send if free margin cannot cover the order. There is no scaling, no martingale, and no grid; the EA's only safety net beyond the regime gate is a per-trade ATR-based stop plus a global daily realized-loss limit of InpDailyLoss (3% of effective capital), a 30-minute cooldown that triggers when ConsecutiveLosses() walks the last seven days of DEAL_ENTRY_OUT history and counts three losses in a row, and a hard cap of InpMaxOpen (3) concurrent positions on the symbol under this magic number.

In-trade management runs every tick. The first line of defense is a break-even move that fires once the position is at 0.5R in profit (half of the 1:2 target). AdjustToStops() makes sure the new stop respects the broker's stops-level, and a one-shot guard means the BE only happens once. At InpTP_R × R (1.6×ATR × 2 = 3.2×ATR from entry), the position is partially closed: 50% of the volume is closed via TryClosePartial_EX18051, with a [P] tag in the comment preventing re-trigger. A forward-only ATR trail at 2.0× ATR(14) starts ratcheting the stop loss after that. Regime change to CHOPPY triggers an immediate full close with reason "REGIME_CHOPPY". The opposite-signal logic recomputes HullC color on each tick and closes any open position whose direction is now opposed by the new Hull color, tagged "OPPOSITE_SIGNAL". Both the close helpers and the modify helper retry up to three times with 100–200ms sleeps on requote, timeout, price-changed, or price-off retcodes. Filling mode is auto-detected by symbol via SetTypeFillingBySymbol().

Execution defaults to InpDryRun=true, which logs intended orders to the journal without sending them, and InpKillSwitch=false. The trade comment is hardcoded as "Psgrowth.com Expert_18051". The expected runtime footprint is roughly four indicator handles (ATR, ADX, Bollinger Bands, and the HTF EMA) plus three inline Hull objects and one inline SuperTrend, which keeps the per-tick cost low enough for M5 use. A backtest on XAUUSD M5 will be dominated by trend-quality segments; flat or choppy weeks can produce long losing streaks that hit the daily loss limit or trip the cooldown, so reviewers should focus on quality-of-trend periods and on the win rate of the partial-close step rather than the raw trade count. Strategy optimization should start from the 2-of-3 confirm and the regime thresholds (ADX 20/25, BB width 0.04/0.10, ATR-percentile 40/60/70) before touching risk or Hull period.