Pipsgrowth EX18090 TrendFollow

Pipsgrowth EX18090 ("KA-Gold Bot") trades XAUUSD with a 3-of-3 trend-following entry that requires a Keltner-channel breakout, an EMA200 regime gate, and a same-bar EMA10 cross through the channel boundary. The EA is bar-driven: OnTick() refreshes the three indicator buffers, runs UpdateOrders() and TrailingStop(), and only calls OpenTrades() when iTime() on bar 0 has changed. There is no tick-by-tick reentry, so entries are deterministic and slippage is bounded to one new bar.

The custom Keltner channel is built in CalculateSignal() from three independent measurements. The middle line is the standard iMA(50) EMA on close. The upper and lower bands are then computed as middle +/- findAvg(50, shift), where findAvg() walks the last 50 completed bars and averages the per-bar (high - low) range. The result is a Keltner variant whose width scales with the symbol's own recent range rather than with a fixed ATR multiplier, which keeps the breakout threshold self-adjusting on gold's volatility cycle. The function evaluates the band twice: once at shift=1 (the closed bar) and once at shift=2 (the bar before that), so two consecutive readings of the channel are available for the cross logic.

A long setup needs all three of these conditions to hold on the same closed bar: (1) iClose(1) sits above the upper Keltner band, meaning the close itself broke out rather than merely tagging the line; (2) iClose(1) is above EMA200(1), filtering the trade to a confirmed long-term trend; and (3) EMA10 crossed through the upper band between bar 2 and bar 1 (EMA10[2] < upper[2] AND EMA10[1] > upper[1]). The third condition is the trigger that distinguishes a real breakout from a slow drift: a fast EMA crossing the channel during the same bar the close is above it is a strong momentum signature. Sells are the mirror: close below the lower band, close below EMA200, and EMA10 crossed through the lower band from above.

Risk brackets are symmetric and 1:1: InpSL_Pips and InpTP_Pips both default to 500 points on gold (a 1:1 reward-to-risk bracket). Both are converted to price through Pips2Double, which the EA derives from the symbol's digits (3 or 5 digit brokers get a 10x multiplier on the _Point unit so that 500 user pips means 5.00 of price on 2-decimal gold). TP can be disabled by setting InpTP_Pips=0, in which case the trade becomes a pure trailing-managed runner; the trailing function checks this flag with if(InpTP_Pips != 0) newTP = m_position.TakeProfit() and otherwise passes 0 to the modify, leaving whatever the broker has stored. With defaults, the bot is a 1:1 bracket system; the trailing only matters once TP is removed or scaled out.

Trailing logic in TrailingStop() uses a one-shot trigger and then a stepped ratchet. Each side has a boolean flag (isSLBuyOrd_Trigger, isSLSellOrd_Trigger) that flips to true the first time Bid - PriceOpen (or PriceOpen - Ask) exceeds ExtTrailingTrigger (300 pips default). After that point, every new bar where profit is greater than ExtTrailingStop + ExtTrailingStep (300 + 100 = 400 pips) calls TryModify_EX18090 to push the SL to Bid - ExtTrailingStop. The compare m_position.StopLoss() < (Bid - (ExtTrailingStop + ExtTrailingStep)) is the one-way ratchet: SL only tightens, never loosens, and each ratchet step is exactly ExtTrailingStep (100 pips) further into profit. The flags reset in UpdateOrders() when the position closes or when an opposite position appears, so a fresh position starts the trailing cycle from zero.

The money model in CalculateVolume() is the only risk dial. When isVolume_Percent=true (the default), lot size is computed as (InpRisk * FreeMargin) / 100000, floored to the nearest multiple of Inpuser_lot (default 0.01). The result is then clamped to LotsMax / LotsMin. With InpRisk=1.0 and a $1,000 account the formula yields a fraction of a standard lot; the floor to Inpuser_lot (0.01) means anything below that snaps up to the broker minimum. This is not a true percent-of-balance risk model in the SL-distance sense — it does not size off the stop distance — it sizes off free margin only. Traders expecting Kelly-style risk should set isVolume_Percent=false and use Inpuser_lot directly.

Three no-trade gates run before the entry call. CheckSpreadAllow() rejects when acSpread (raw point spread, not pips) exceeds InpMax_spread=65; on a 2-decimal gold feed that allows roughly 6.5 pips, which is permissive for M5 trend-following. CheckVolumeValue() validates the lot against SYMBOL_VOLUME_MIN, SYMBOL_VOLUME_MAX, and the volume step. CheckMoneyForTrade() runs OrderCalcMargin to ensure the lot has enough free margin for a hypothetical fill. CheckStopLoss() then confirms that |price - SL| and |price - TP| are both outside the broker's SYMBOL_TRADE_STOPS_LEVEL, which avoids the TRADE_RETCODE_INVALID_STOPS rejection on brokers with a freeze zone. The InpTimeFilter, when true, restricts entries to 02:30 - 21:00 server time — that window covers Asia-open through the NY close and is intentionally broad; the only days excluded are the rollover gap (22:00-02:30) and weekends.

The execution helpers TryClose_EX18090, TryClosePartial_EX18090, and TryModify_EX18090 are wired as standard 3-attempt wrappers that retry only on REQUOTE / TIMEOUT / PRICE_OFF / PRICE_CHANGED with a 200ms sleep (100ms for modify) and break out on any other retcode. The partial-close and close helpers exist but are not actually called from the EA's main flow — OpenTrades() uses m_trade.Buy / m_trade.Sell directly with a single attempt, and TrailingStop() routes every modify through TryModify_EX18090. So in practice the resilience layer is asymmetric: modifications are retried, entries are not. On a noisy news tick that whiffs the entry, the trade is simply missed on that bar and the next bar gets a fresh attempt.

A note on the source: the SELL branch in CalculateSignal() reads result = true rather than result = ORDER_TYPE_SELL, which compiles to the integer 1 — and ORDER_TYPE_SELL is also 1 in ENUM_ORDER_TYPE, so the EA works by accident. If a future refactor changes the enum values, the sell side would silently start emitting order types that no condition ever reaches. This is a code-smell worth flagging if you ever fork the bot.

On a strategy-style summary: this is a momentum-continuation, not a mean-reversion, EA. The Keltner+EMA10 cross is a high-conviction breakout pattern that fires rarely — typically a handful of times per week on XAUUSD M5 — and the 1:1 bracket plus stepped trail is built to let winners run when TP is disabled and to lock quickly when TP is on. Set InpTP_Pips=0 and the trailing takes over; keep InpTP_Pips=500 for a faster, bracket-only behavior. The EMA200 filter alone rules out roughly half the candles on gold's chop regime, so the win rate is the function of gold's trending periods, not the EA's logic. Backtests will look best across sustained trends (2020 summer, 2022 Q1, 2024 spring) and worst during the tight $30 ranges that print 50+ false breakouts in a row.

For live deployment: any ECN/STP broker with raw spreads under 50 points on XAUUSD will keep CheckSpreadAllow() open most of the time. The 500/500 point bracket is wide enough that micro-lot accounts ($100-300) are workable, though slippage on the entry will be a larger percentage of profit. Run on a VPS with sub-100ms broker latency for the trailing to actually move inside the same bar as the trigger; on a home connection the 300-pip trigger ratchet can be missed by a full bar, which changes the geometry of the trade.