Pipsgrowth EX18002 TrendFollow

EX18002 Ichi_ATR_EA is a single-position trend-following system that reads the Ichimoku Kinko Hyo cloud on the working timeframe and confirms every signal with a higher-timeframe EMA, a multi-input regime classifier, and an ATR-derived risk envelope. The strategy is not a generic 'open on cross' rule. The entry code in ComputeSignal() rejects a Tenkan/Kijun crossover unless the closed bar's close price sits on the correct side of the projected kumo (Senkou Span A vs Senkou Span B at the closed bar). A bullish cross that prints inside or below the cloud returns no signal; a bearish cross above the cloud is ignored. When both conditions align, the entry confidence is 75; a cross against the cloud still produces a direction but at 50, so the engine prefers cloud-aligned setups when it can.

The regime layer in ComputeRegime() classifies the market into seven states (STRONG_TREND, WEAK_TREND, RANGE, BREAKOUT, COMPRESS, EXPAND, CHOPPY) using three signals read from the ADX(14), ATR(14), and Bollinger Bands(20, 2.0): the absolute ADX level, the current ATR's percentile rank over the prior 100 bars, and whether the Bollinger band width is expanding. STRONG_TREND fires when ADX crosses 30; WEAK_TREND uses the configurable InpADXThreshold (20.0 default). COMPRESS triggers when Bollinger width falls below 4% of midline and is not expanding. BREAKOUT and EXPAND fire when ATR-percentile is above 60% and bands are widening. CHOPPY locks entries out entirely when ADX drops below 15. Only four regime states allow new entries: STRONG_TREND, WEAK_TREND, BREAKOUT, and EXPAND — the others short-circuit TryEntry().

The confirm step in CheckConfirm() reads a higher-timeframe EMA(50) on the InpHTF timeframe (default index 8, mapping to H1) and refuses any entry where the higher-TF close is on the wrong side of that EMA. The same function enforces a minimum reward-to-risk ratio through InpMinRR (1.5 default): the engine computes the SL distance as ATR × 1.5 and the TP distance as ATR × 2.5, and if the resulting RR falls below 1.5 the signal is discarded. The position-size call in CalcLots() then sizes the trade off effective_capital, not raw equity: when InpCapAmount is set above zero, effective capital is the lesser of the configured cap and the live account equity. Below InpCapFloor (50.0) the engine refuses to open anything at all.

The no-trade gate in CheckNoTrade() runs before every entry. It enforces a weekend block (Saturday/Sunday), a 7-to-21 server-time window (the London-through-NY active session), a rolling-spread filter that compares the current spread to a 50-bar mean and rejects it once the multiple exceeds 3.0×, a daily-loss cap of 3% of effective capital, a weekly-loss cap of 6%, the InpMaxConcurrent ceiling (1 default), a kill switch, and a hard broker check (SYMBOL_TRADE_MODE_DISABLED). OnTradeTransaction() tracks consecutive losses; once g_consecLosses reaches InpCooldownLosses (3 default) it sets a 10-bar cooldown timer and clears the counter.

Management runs on every tick, not just on a new bar. ManageOpenPositions() runs five exit paths in sequence. Exit 1 closes the trade when the current signal flips to the opposite direction. Exit 2 closes after 200 bars in trade (MAX_BARS_IN_TRADE). Exit 3 takes a 50% partial close at +1R profit, leaving the remainder to trail. Exit 4 moves the stop to breakeven — slightly above entry on buys, slightly below on sells — once price travels +1R. Exit 5 then engages a pure ATR trail: as price moves in profit, the stop follows at ATR × InpTrailATRFactor (2.0 default), and only ratchets forward (never backward). The order-execution path in SendOrder() does a margin pre-check, respects the broker's stops-level and freeze-level distances, and retries once on TRADE_RETCODE_REQUOTE, TRADE_RETCODE_TIMEOUT, or TRADE_RETCODE_PRICE_CHANGED after a 200ms sleep. Modify, close, and partial-close helpers (TryModify_EX18002, TryClose_EX18002, TryClosePartial_EX18002) each do up to three attempts on requote/timeout before giving up.

The risk model is fixed-fractional at 0.5% of effective capital per trade by default (InpRiskPercent). With a 1.5×ATR stop on XAUUSD M5, that risk budget maps to a small lot fraction on a $1,000 account. The combined effect of the 3% daily loss cap, 6% weekly cap, 200-bar time exit, and BE+trail on the open side keeps the worst-case per-trade damage bounded: a single adverse position is closed either by stop, by time, or by an opposite signal before it can compound into a larger drawdown. The OnTester() formula is (netProfit × profitFactor) / (1 + equityDDPct), and it requires at least 30 trades before it returns a nonzero score, so the optimizer will not reward a backtest that produced too few fills to be meaningful.

Because the strategy holds one position at a time and respects a hard 7-to-21 session window, the typical exposure profile is a small handful of trades per week on the M5 default and slightly more on M15. InpDryRun is true by default, so the EA logs intended orders to the journal without sending them — flip it to false once the live account, magic, and capital cap are configured for the symbol you intend to trade.