Pipsgrowth EX18027 TrendFollow

Pipsgrowth EX18027 MultiIndicator TrendFollow is a per-bar consensus EA that fires only when at least three of four inlined trend primitives agree on direction. The four primitives are not stock MQL indicators — they are rebuilt in code from EMA and price primitives during the same bar. The first is EN-Candle-Color: a binary vote on whether the most recent closed bar printed a bullish body (close[1] > open[1]) or a bearish one. The second is EN-zero-lag-T3, the Tillson T3 smoothed moving average: ComputeT3() walks the last up-to-200 bars through six nested EMAs of period InpFastPeriod=14, then weights those six EMA outputs with the four Tillson coefficients derived from InpT3Hot=0.7 (c1 = -v³ = -0.343, c2 = 3v² + 3v³ = 1.911, c3 = -6v² - 3v - 3v³ = -3.207, c4 = 1 + 3v + 3v² + v³ = 3.639), producing a T3 line that lags price by far less than a stock EMA of the same period. The T3 vote is bullish when close[1] is above the T3 line, bearish when it is below. The third is EN-zero-lag-MA, the standard ZLEMA constructed inside ComputeZLMA() as EMA(2*close[i] - close[i+period], period) — that lag-correction trick cancels most of the EMA's lag, so the ZLEMA vote is bullish when close[1] is above the ZLMA line AND the ZLMA line is rising (ZLMA[1] > ZLMA[2]). The fourth is EN-ultra-trend-zero-lag-MA: a simple comparison of the fast ZLMA(14) against the slow ZLMA(50) — bullish when fast ZLMA is above slow ZLMA, bearish when it is below. ComputeSignal() increments the buy counter on each bullish primitive and the sell counter on each bearish one, then assigns a confidence score of |buyVotes - sellVotes| / 4 * 100. A trade only opens when one side has InpVoteThreshold=3 votes and beats the other side outright; the third bar in a clean trend almost always delivers 3 or 4 votes while a chop bar rarely delivers more than 2.

The signal then has to pass three layers of gating before TryEntry() will fire. The first is the regime classifier in DetectRegime(). Each closed bar, the EA computes ADX(14), the Bollinger(20, 2) width-as-fraction-of-middle, and the ATR(14) percentile over a 50-bar window. Seven states are possible: REG_STRONG_TREND requires ADX>=30 and ATR-percentile above 70%; REG_WEAK_TREND requires ADX>=22; REG_EXPAND requires Bollinger width above 3% of the middle AND ATR-percentile above 60%; REG_COMPRESS requires Bollinger width below 1% AND ATR-percentile below 30%; REG_BREAKOUT requires ATR-percentile above 85% AND Bollinger width above 3%; REG_CHOPPY triggers when ADX drops below 15; everything else falls into REG_RANGE. Choppy is the only regime the EA hard-blocks via CanTradeNow(); Compress, Range, and Breakout are all eligible for entries because each contains trend moments that the four-vote consensus can pick up.

The second gate is the higher-timeframe confirmation inside TryEntry() itself. The EA reads an EMA(InpHtfEmaPeriod=50) from the higher timeframe selected by InpHtfTF. The InpHtfTF input is a 1-7 mapped through MapTimeframeInt() (1=M1, 2=M5, 3=M15, 4=M30, 5=H1, 6=H4, 7=D1), and any value outside that 1-7 window — including the EA's default InpHtfTF=8 — falls through the switch to PERIOD_H1. So the default configuration confirms against the H1 50-EMA: a buy requires ask >= htfEma(50) at H1, a sell requires bid <= htfEma(50). This filter alone is what keeps the EA out of M5 trades that are running counter to the H1 trend.

The third gate is CanTradeNow(). It checks: (1) InpKillSwitch is off; (2) the terminal and account both allow trading; (3) symbol rates refreshed successfully; (4) live spread is below InpMaxSpreadPoints=30 points; (5) live spread in price terms is below 50% of the ATR(14) — so a 30-point spread on a 60-point ATR(14) is fine but a 30-point spread on a 40-point ATR(14) is not; (6) the server clock is in the 7..21 window (no weekends, no low-liquidity hours); (7) when InpCapitalCapAmount>0 and g_effectiveCapital drops below InpCapitalCapFloor=$50, no new entries — the default cap is 0 (cap off), but the $50 floor still blocks under-funded accounts; (8) the day's realized PnL hasn't already lost InpDailyLossLimitPct=3% of g_effectiveCapital; (9) the week's realized PnL hasn't already lost InpWeeklyLossLimitPct=6%; (10) the cooldown gate has not been engaged by InpCooldownLosses=3 consecutive losses (which then locks the EA out for InpCooldownBars=20 closed bars); and (11) the regime is not REG_CHOPPY.

Position management runs every tick from ManageOpenPositions(). The order in which the EA processes a position is: time-exit first (if BarsInTrade(openTime) >= InpMaxBarsInTrade=200, close with reason TIME_EXIT), then break-even (if R-multiple >= InpBreakEvenR=1.0 and the current stop is still below entry, snap SL to entry+2 points and continue), then partial close at TP1 (if InpPartialPct=50 and bid/ask has reached open+0.5InpAtrTpMultATR for buys, or open-0.5InpAtrTpMultATR for sells, and the remaining lot after partial is at or above the broker minimum, close half), then ATR trailing (snap SL to bid-trail for buys, ask+trail for sells, where trail = InpAtrTrailMult=2.5 * ATR(14), and only ratchet in the favorable direction). Each modification retries three times on requote, timeout, price-off, or price-changed before giving up. The order matters: if time-exit fires, the position is gone and break-even never runs; if break-even moves the SL, the trailing pass then re-evaluates whether the new SL is still the right distance behind price.

Initial SL is set to InpAtrSlMult=1.8 * ATR(14) on the bar the trade opens; initial TP is set to InpAtrTpMult=3.0 * ATR(14), which works out to a 1:1.67 reward-to-risk ratio against the SL. The EA rejects any candidate trade where the projected TP distance is smaller than the projected SL distance (the ratio must be at least 1.0). Sizing goes through CalcLotSize(): riskMoney = g_effectiveCapital * InpRiskPercent/100 = effective_capital * 0.005, divided by per-lot stop-loss value (slDist / tickSize * tickValue), then floored to the symbol's volume step and clamped to the symbol's volume min/max. Before any order is sent, OrderCalcMargin() verifies that the requested lot × entry price requires less than the account's free margin; if not, the order is refused without consuming a ticket attempt. The execution helper TryEntry() itself has a 1-retry loop for requote/timeout/price-changed before the trade is given up on for the bar. The InpDryRun=true default means the EA logs every decision to the journal (with votes, confidence, regime, lots, slDist, tpDist) but does not place orders until you flip that input off.

The original-source design of EX18027 is best understood as a vote-then-confirm model. The 3-of-4 threshold (InpVoteThreshold=3) is intentionally stricter than 2-of-4: a single bar in a 50% chop environment can produce 2 buy votes and 2 sell votes, but a clean trend almost always produces 3 or 4 votes on the dominant side within one to two bars. Pairing that with the H1 50-EMA confirmation and the 1:1.67 R:R means the EA is a conservative swing-on-5min EA: it does not take many trades, it waits for H1 to agree, it manages each trade through break-even at 1R, partial at 1.5R, and ATR trail from there, and it bails after 200 M5 bars (about 16 hours) if the trade hasn't moved. OnTester() multiplies net profit by profit factor and divides by 1+balance-drawdown, with a 30-trade minimum before the score is non-zero — a metaheuristic that lets the strategy tester rank passes that have statistically meaningful sample sizes.

A practical note on the InpHtfTF=8 default: any value outside 1..7 silently maps to H1, so the default is in fact the H1 confirmation, but if you set the input to 4 expecting M30 confirmation you have to know to enter 4 not 8. If you want pure intraday confirmation, set InpHtfTF=5 (H1) explicitly; if you want daily-trend confirmation, set it to 7 (D1); if you want the EA to trade within a H4 swing, set it to 6. The H1 50-EMA is the well-tested default and the only configuration the EA's published backtest covers. Other choices (H4 50, D1 50) are valid but represent parameter exploration the EA does not pre-tune for you.