Pipsgrowth EX01002 Adaptive

EX01002 Adaptive runs a regime-dispatched moving-average crossover. The core idea is that the same MA crossover cannot be trusted in every market, so before any signal fires the EA first asks ADX(14) what kind of market it is looking at, then lets the same close-vs-AMA test behave differently depending on the answer. The dispatcher is ClassifyRegime() and the only signal generator is MasterSignal().

The regime classifier reads two values on the working timeframe: ADX(14) and ATR(14). When ADX is at or above InpRegime_ADX_Trend (25 by default) the regime is STRONG_TREND; between InpRegime_ADX_Range (20) and 25 it is WEAK_TREND; below 20 it is RANGE. The classifier also looks at ATR in pips and blocks the entire trade when ATR is below InpATRMinPips (1 pip) — the spread would eat the move — or above InpATRMaxPips (0 = disabled) when volatility is too high for the fixed stop distance. Three more enum values (NEUTRAL, BREAKOUT, CHOPPY) exist in the type definition but are not produced by ClassifyRegime() in the current build; they are reserved for the broader Adaptive family.

MasterSignal() is the only place the EA produces a directional opinion. It reads two prior AMA buffer values (the 20-period SMA on close, default) and the close of the two most recent closed bars. In STRONG_TREND and WEAK_TREND, a buy fires either on a strict close[2]<AMA_prev and close[1]>AMA_curr crossover or on a continuation pattern where both closes sit above the AMA in the direction of the trade. In RANGE, NEUTRAL, and BREAKOUT, only the strict crossover is accepted. The continuation fallback is suppressed on purpose in ranging markets because in a range, two closes above the MA usually means a fake breakout rather than a real move.

The third gate is higher-timeframe alignment. HTFTrendAgrees() copies the close of bar 1 on a higher timeframe (default H4, set with InpHTF_TF=8) and compares it to a 50-period EMA on that same higher timeframe. A long signal on the working timeframe is allowed only if HTF close is above HTF EMA; a short only if HTF close is below HTF EMA. The HTF can be reconfigured from M1 to D1, and the EMA period from 5 to 500.

The no-trade gate is the longest single function in the file. NoTradeGate() runs nine checks in sequence and returns true (gate closed) at the first failure. It checks market open (Friday after 22:00 GMT blocked, Sunday before 22:00 GMT blocked), active session (London 08:00–16:30 GMT and New York 13:00–21:00 GMT; Asia 22:00–06:00 blocked by default via InpAvoidAsia), spread against InpMaxSpread (50 points default), effective capital against InpCapFloor (50 if a cap is set), trades today against InpMaxDailyTrades (5), realized today against -5% of effective capital, realized week against -9% of effective capital, cooldown timer (30 minutes after any losing close), consecutive losses (max 3 by default), total drawdown from session start against 20% (which latches g_ddKillSwitch and blocks all new entries), open-position cap (3 default), and the pyramid gate. The g_ddKillSwitch is a hard latch — once tripped, the EA refuses every new entry until the terminal restarts or the EA is reinitialized.

Lot sizing is risk-based. CalcLotByRisk() takes 1% of effective capital (the lower of InpCapAmount and current equity) and divides by the per-point SL cost, computed as StopLossPips × point × tickValue / tickSize. The resulting lot is rounded to the broker's lot step, clamped to the symbol's min and max, and capped at InpMaxLot (100 lots). When RiskPerTradePercent is set to zero the EA falls back to FixedLotSize=0.01 — useful for prop-firm challenge accounts where percentage risk is restricted.

Order sending goes through a wrapper with three retries. TryBuy_EX01002 / TrySell_EX01002 is called, and on requote, timeout, or price-changed retcodes the wrapper sleeps 100ms, refreshes the symbol, and retries. The wrapper also clamps the SL to the broker's minimum stop distance (SYMBOL_TRADE_STOPS_LEVEL) to prevent rejection, and it pre-checks free margin — if the order would exceed free margin, the entry is skipped without retries. The minimum R:R is enforced at the confirm step: TakeProfitPips/StopLossPips must be ≥ InpMin_RR_Ratio (1.2 default), so the default 1000/500 = 2.0 R:R passes comfortably.

ManageTrades() runs every tick. It walks the open positions for this magic and symbol, and applies in order: time-based exit at InpMax_Hours_In_Trade (240 hours = 10 days default), opposite-signal exit (close if MasterSignal() flips against the open direction), partial TP at 50% of position when profit reaches TakeProfitPips × InpPartial_TP_Ratio (i.e. 500 points with the default TP of 1000), trailing stop (200 points, ratchets only in the favorable direction), and the optional break-even (off by default; when enabled, SL moves to entry + 2 points once profit reaches 100 points) and profit lock (off by default; locks SL 20 points behind price for every 50 points of profit gained). The partial-close is tracked in a 100-slot array of ticket/partial-done flags so the same position never partials twice.

The trailing logic is a single-direction ratchet. For a long, once price has moved 200 points in profit, the new SL is current_price - 200, but only if that is higher than the current SL and above the open price. The same logic applies in mirror to shorts. Because the check runs every tick, the ratchet can step multiple times within a single candle when price is moving fast.

Cooldown and consecutive-loss tracking is event-driven. OnTradeTransaction() listens for DEAL_ADD events matching the magic and symbol. On entry deals, tradesToday increments. On exit deals, the net P&L (profit + swap + commission) is added to realizedToday and realizedWeek. When the net is negative, consecLosses increments and cooldownUntil is set to now + 30 minutes. When the net is positive, consecLosses resets to zero. The cooldown is a hard gate — the EA will not consider a new entry until the timer expires, regardless of how strong the signal looks.

The daily and weekly loss caps use effective capital as the denominator. With InpDailyLossPct=5.0 and effective capital of $1,000, the EA halts new entries once realizedToday drops to -$50. The realized totals reset at the start of each trading day and each Monday (in broker-local time, then converted to GMT via DetectGMTOffset() which reads the weekend gap in the H1 bar history).

What this EA does not do matters as much as what it does. There is no martingale, no grid, no averaging down. The only scaling mode is the pyramid gate, and PyramidGate_ProfitOnly / PyramidGate_ProfitAndSL both require every prior position to be in profit before any new entry is allowed. There is no news filter, no hedging, no swap-arbitrage. The flat zone between ADX 20 and ADX 25 is where the strategy's hit rate will degrade, and that is exactly why the daily/weekly/drawdown caps exist as the second layer of safety.

A reasonable backtest plan: run the MT5 Strategy Tester on XAUUSD M5 with "Every tick based on real ticks" for at least 12 months, and lower InpMaxDDPct to 10% during the test to confirm the kill switch actually trips. If the realized drawdown in the backtest is consistently below the configured cap, the safety layer is wired correctly.