Pipsgrowth EX18045 TrendFollow

Pipsgrowth EX18045 (TrendFollow v2.00, magic 22218045) is built around a stripped-down HeikinAshi colour-flip engine. It does not pull a HeikinAshi buffer from any built-in indicator; instead, the OnTick handler rebuilds the entire HA candle stack bar by bar from the raw MQL5 time-series calls (CopyOpen / CopyHigh / CopyLow / CopyClose), then watches the colour transition between the second-to-last closed HA candle and the most recently closed HA candle. A red candle that becomes green returns sig = +1, a green candle that becomes red returns sig = -1, and a same-colour pair returns 0. The base confidence stamp is 55, then the function computes body-to-range ratio on the closed signal bar (MathAbs(haClose[1] - haOpen[1]) divided by the highest-high minus lowest-low of that same bar) and adds that ratio scaled by 45, so a strong full-body candle pushes confidence to 100 and a doji-like candle leaves it at 55. No moving average, oscillator, or volatility measurement is read for the signal itself — colour flip plus body strength is the entire entry thesis.

That simplicity is then buried under a thick regime filter. DetectRegime() reads ADX(14), ATR(14) and Bollinger Bands(20, 2.0) and classifies the market into seven possible labels. StrongTrend requires ADX >= 25 and ATR-percentile above 0.7 (the percentile is computed against the most recent 50 ATR values, not a fixed threshold); WeakTrend is the same ADX gate without the volatility confirm. Compress triggers when the percentile drops below 0.25 and Bollinger width is positive; Expand is the symmetric mirror above 0.75. Breakout requires ATR-percentile above 0.6 plus ADX above 18, Choppy triggers when ADX drops below 15, and everything else is Range. OnTick only accepts entries when the regime is StrongTrend, WeakTrend or Breakout — and even in those regimes the call still has to clear two confirm gates inside ConfirmSignal() before any size is calculated. The first gate checks that the current Bid is on the correct side of the H1 EMA(50) for the proposed direction (long needs price above, short needs price below); the second gate checks that the actual M5 candle closed in the right direction (close > open for long, close < open for short). A failed confirm prints the reason — 'HTF trend not up', 'signal bar not bullish close' — and the cycle ends without a position.

The no-trade gate in CanTrade() is unusually busy. It checks the kill-switch flag, walks the realised-today/realised-week PnL history for this magic against the daily 3% and weekly 9% loss caps, enforces the 30-minute cooldown that OnTradeTransaction arms after three consecutive losses, refuses trades when the broker's current spread exceeds InpMaxSpreadPoints = 50 points, refuses when total positions for the EA have reached 5, refuses when symbol exposure has reached 3, blocks Saturdays, Sundays and Friday after 22:00 server time, blocks a 15-minute news window around 12:30 server time, a 10-minute window around 13:55, and a 5-minute window after 14:00, and finally blocks the 23:00 to 01:00 server-time window for low liquidity. The news-window and low-liquidity blocks are unusual inside this family — most siblings let the broker hours do the filtering.

Position math is conventional for the family. SL distance is 1.5×ATR(14), TP distance is 3.0×ATR(14), giving a fixed 1:2 reward-to-risk ratio that ComputeLevels() also validates against InpRRMin = 1.5. Lot sizing in ComputeLotSize() takes 0.5% of the EffectiveCapital (which is min(account equity, InpCapitalCapAmount) when the cap is on, otherwise raw equity, with a $50 floor below which no new entries are allowed), divides it by the per-lot loss implied by the SL distance in points, and clamps the result to the broker's volume step / min / max and the InpMaxLotSize = 10 ceiling. OpenTrade() calls OrderCalcMargin to pre-check that the account's free margin covers the new order, then sends with m_trade.Buy / m_trade.Sell using the auto-detected filling policy and a 20-point deviation. There is a single retry on requote, timeout or price-off, refreshing the symbol rates first.

ManageOpenPositions() runs every tick and walks the positions owned by this magic. The first check is the 60-bar time stop (InpMaxBarsInTrade = 60 is much shorter than the 200- to 288-bar exits seen on siblings in the same family). The second check force-closes if the regime has rolled into Choppy or Range. The third check is a volatility-spike exit that closes the trade if the current ATR is more than 2.5× the ATR observed at entry — a stop-loss on a runaway volatility event, distinct from the standard opposite-signal exit. After those gates, the function evaluates the live R-multiple; when rMultiple >= InpBreakevenAtR = 1.0 it ratchets the stop to the entry price (one-shot, no further BE adjustments). At the half-distance to the full TP, the partial close fires: it attempts to close InpPartialTPPct = 50% of the position, but only if the remaining volume stays above the broker's minimum lot. Finally, the ATR trail at 2.5×ATR(14) is applied forward-only — for longs, the new stop is the current bid minus the trail distance, and it only replaces the existing stop if it is higher than the current stop and above the entry price. The function never lowers a stop in either direction.

A few design choices deserve to be called out for traders sizing this up. First, the signal is intentionally minimal — a single HeikinAshi flip — so the EA will take entries in a StrongTrend or Breakout regime that other EAs in the family would not have entered because they require three or four voting indicators to align. Expect more trades, smaller individual edge. Second, the vol-spike exit means the position can close in loss during a fast spike even if the stop is not yet hit, because the entry-time ATR is the reference — a sudden 2.5× volatility event is treated as a structural break, not as a stop-hunt. Third, the 30-minute cooldown after three consecutive losses is much shorter than the 4-hour lockout some siblings use, so the EA is more willing to re-engage after a drawdown streak. Fourth, the 9% weekly loss cap is one of the looser caps in the family — it accepts a larger drawdown envelope in exchange for not blocking the strategy during its recovery. Finally, the default InpDryRun is true, so the EA logs every entry, partial, modify and close as a DRYRUN line to the journal until the user flips the flag to false. The user has to consciously go live.