Pipsgrowth EX18050 TrendFollow

Pipsgrowth EX18050 TrendFollow is a single-position M5 trend follower built around an inlined Tillson T3 moving average applied to a Heikin-Ashi-smoothed close. Rather than a fast/slow EMA cross or a MACD histogram flip, the entry engine watches the slope of a T3 line that the EA itself computes from six nested EMAs of period 14, weighted by Tillson's volume factor of 0.7. The T3 is drawn on top of a four-price HA close rather than raw candles, so the curve stays smooth through sideways noise and only bends meaningfully when the underlying rhythm shifts. A long signal fires when the current T3 value sits at least five points above its prior-bar reading and the actual close also sits above the previous T3 value; a short signal mirrors that pair. Because both legs of the test must clear, the EA refuses to enter on a slope that is rising into price still trading under the curve, and vice versa — that is the repaint guard that makes the strategy usable on live data, not just backtested curves.

The signal lives inside a seven-state regime classifier in ComputeRegime(). The classifier uses an ADX(14) reading, the 50-bar percentile of ATR(14), and the ratio of current Bollinger Band(20, 2.0) width to its prior bar. STRONG_TREND requires ADX above 22 and an ATR percentile at or above 60 — momentum plus volatility expansion in agreement. WEAK_TREND triggers when ADX clears 15.4 but the volatility percentile hasn't followed. EXPAND is detected by a 20%+ widening of the Bollinger envelope even when ADX is still building, while COMPRESS flags the opposite. BREAKOUT and RANGE are pure ATR-percentile buckets (>=75 and <=25 respectively). Anything that doesn't fit lands in CHOPPY, which the strategy treats as a hard block. The RegimeAllowsEntry() gate then narrows the tradeable set to STRONG_TREND, WEAK_TREND, BREAKOUT, and EXPAND, which is why the EA sits out of compression and pure range regimes entirely rather than getting chopped up at the band edges.

Confirmation is a three-gate panel. First, the higher-timeframe EMA(50) — pulled on the timeframe set by InpHTF (default H1) — must slope in the same direction as the signal; a long signal against a falling H1 EMA is refused. Second, current spread must sit at or below InpSpreadMaxPts of 30 points. Third, the active session must match one of three booleans: London 08:00–16:00 server, New York 13:00–21:00 server, or Asian 00:00–08:00 server. An RSI(14) extreme-check sits behind those gates: longs are refused when RSI is above 80, shorts when below 20. The CheckConfirm() helper returns the first failure reason as a string so the EA can log exactly which gate blocked.

The risk engine sizes every entry off the signal ATR — iATR on the signal timeframe with period equal to the T3 period (14). Stop loss is InpATRStopMult of 2.0× that ATR, take profit is InpATRTargetMult of 4.0× ATR, hardcoding a 1:2 reward-to-risk ratio. If a calculation slips below the InpMinRR of 1.5, the target is widened until the ratio clears. Position size is then computed in PreSendRiskOK() against InpRiskPercent of 0.5% of effective capital (or capped at the lot ceiling), with OrderCalcMargin checking free margin before the order is dispatched. The InpCapAmount input sets an optional equity ceiling — set to 0 it disables the cap; InpCapFloor of $50 blocks entries once effective capital drops below that level. A hard cap of 50 trades per day is enforced through _MAX_DAILY_TRADES_HARD.

Once in a position, ManageExits() runs a layered exit stack on every tick. When unrealized profit reaches one R multiple, MoveSL() slides the stop to entry plus the InpBEOffsetPts of 10 points, locking in a scratch-free exit. At the same 1R mark, half the position is closed via TryClosePartial_EX18050 (three attempts on requote, timeout, price-off, or price-changed, with 200ms sleeps between tries). Beyond 1R, a 2.5×ATR trailing stop takes over and only ever moves in the favorable direction. Two hard exits run in parallel: an opposite T3-signal exit closes the trade, and a regime flip to CHOPPY liquidates the position regardless of profit. A 500-bar time-based stop finally kicks in if the trade has been alive that long on the entry timeframe — a generous cap designed for the slow XAUUSD trend this EA targets.

A loss-cooldown watchdog sits in OnTradeTransaction. Every deal tagged with the EA's magic number is read; three consecutive losses activate a four-hour cooldown during which NoTradeReasons() returns the string "cooldown after losses". Daily and weekly equity-anchored loss limits (3% / 6% default) similarly pause entries once breached. A duplicate-entry guard keyed on the signal-bar timestamp prevents the EA from firing twice on the same close. Pyramiding is compiled in via MaybePyramid() and capped at _PYRAMID_MAX_LEVELS of 3 with 1.5×ATR spacing, but the InpPyramidEnable input is false by default — out of the box, EX18050 runs one position at a time. The InpDryRun input defaults to true, so the EA prints intended orders to the log without sending them; the user has to flip that to false to trade live. InpKillSwitch provides a manual panic button that hard-stops new entries from the next tick.

The OnTester fitness function is (netProfit * profitFactor) / (1 + drawdownPct/100), gated on a 30-trade minimum and a positive PF. That makes the optimizer reward raw return dampened by drawdown rather than chasing profit alone, which is why a parameter set with a smooth equity curve and modest profit will outrank a spiky high-PF one in optimization.

In practice, expect EX18050 to take a handful of trades per week on XAUUSD M5 during active sessions, with most winners carried by the trailing stop rather than the fixed 1:2 target. Backtest curves on gold typically show long flat stretches punctuated by 2:1 to 4:1 winners when a trend extends. The InpDryRun=true default means the EA is safe to drop on a chart immediately for visual verification — the user toggles the live switch only after confirming the signal and regime logic against their broker's data.