Pipsgrowth EX18115 TrendFollow

EX18115 HedgingMartingale is, despite the marketing header that advertises a 'full 12-layer stack' of regime detection, signal generation, confirmations, no-trade gates, capital caps, risk engines, scaling, and an OnTester pass, in practice a single-file recovery script. The header at the top of the .mq5 lists twelve architectural layers (REGIME, SIGNAL, ENTRY, CONFIRM, NO-TRADE, CAPITAL CAP, RISK, SIZING, MANAGE, EXIT, SCALING, OnTester), but if you open the file you will find only six functions — BasketProfit, CountOpen, CloseAll, and three retry wrappers TryClose_EX18115, TryClosePartial_EX18115, TryModify_EX18115 — plus a single OnTick() body that fits on twelve lines. There is no regime classifier, no indicator handle, no signal function, no confirmation gate, no news filter, no session filter, no OnTester, no magic-number check beyond the per-position filter inside the helpers, and no dashboard or chart object. The header is aspirational copy, not a description of the code. Anyone reading this article should treat the .mq5 itself as the source of truth and the header as marketing language.

What the EA actually does is one specific, well-known recovery pattern. When OnTick() finds zero open positions and the initialOpen flag is false, it reads the current ASK price and calls trade.Buy() with InpInitialLot (default 0.10) and the comment Psgrowth.com Expert_18115. The stop loss and take profit passed to trade.Buy() are both 0 — there is no per-trade protective stop on the entry. The flag initialOpen flips to true and the EA waits. If the price moves against this initial long by enough points, the basket as a whole goes negative; the EA then layers in a SELL hedge to lock the drawdown into a closed lot-multiplication cycle.

The hedging leg is governed by three inputs that operate together. InpEnableMartingale (default false) must be true for the recovery to run. InpRecoveryStep (default 100, in points) sets the per-level drawdown trigger in raw points, and the trigger is scaled by (level+1), so the second hedge fires when the basket is down 100 * 2 = 200 points, the third at 100 * 3 = 300 points, and the fourth at 100 * 4 = 400 points. InpHedgeLot (default 0.20) is the base lot for the first hedge, and the actual hedge lot is NormalizeDouble(InpHedgeLot * MathPow(2, level), 2) — so the levels are 0.20, 0.40, 0.80, 1.60 lots on gold-sized accounts by default. InpMaxLevels caps how many hedges can be opened in one recovery cycle (default 4). Once that cap is hit, the EA simply does not add any more positions and waits for the basket to come back. There is no breaker that hard-stops the EA at a portfolio drawdown threshold; the only safety is the InpMaxLevels cap.

The exit logic is the simplest part of the design and the only condition that actually closes the basket. On every tick, after recomputing BasketProfit() and CountOpen(), the code checks if(cnt>0 && basket>=50). If the basket is positive by 50 account-currency units (no scaling by lot, no scaling by step, a flat $50 / €50 / ¥50 / XAU 50 etc. target), CloseAll() is called, which iterates PositionsTotal() from the highest index down, filters on the magic number 22218115 and the chart symbol, and sends a close through TryClose_EX18115. That helper retries the close up to three times with a 200 ms Sleep() between attempts on requote, timeout, price-off, or price-changed return codes, and returns false on the third failure. After CloseAll() runs, the EA resets level = 0 and initialOpen = false, so the next tick with no open positions will start a fresh cycle with a fresh InpInitialLot buy. There is no partial close, no per-leg TP, and no trailing stop.

A few honest observations about the implementation. The trade.Buy() and trade.Sell() calls both pass 0, 0 for stop loss and take profit, so the only thing standing between the account and an unbounded drawdown is the InpMaxLevels cap and the broker's margin call. There is no equity-curve stop, no daily-loss stop, no per-leg SL, and no kill-switch. The OnTick() body has a return after the first Buy and another return after the basket-profit close, so the rest of the function is skipped on those ticks; in practice the EA processes at most one decision per tick. The retry wrappers TryClosePartial_EX18115 and TryModify_EX18115 are declared and compiled, but nothing in the active path calls them — CloseAll() only calls TryClose_EX18115, and there are no modifications to existing positions anywhere in the file.

The magic number is hardcoded as 22218115 in the InpMagicNumber input and is also applied to the CTrade instance via trade.SetExpertMagicNumber(InpMagicNumber) in OnInit(). The trade comment is the literal string Psgrowth.com Expert_18115 for the initial buy and Psgrowth.com Expert_18115 Lvl <n> for hedges, which makes it easy to identify the EA's positions in the trade panel and in any third-party reporting tool. The fill mode is whatever the broker advertises — the EA never calls trade.SetTypeFilling() and never overrides ORDER_FILLING_*, so on ECN/STP accounts the broker's policy applies automatically. Slippage tolerance is set with trade.SetDeviationInPoints(30), which is generous for gold on most retail brokers but can still get clipped on fast news spikes.

In backtest, this EA will print a small handful of trades per session on M5 XAUUSD. The equity curve will be stair-stepped: long flat segments interrupted by sudden multi-leg drawdowns that resolve in a single basket profit close, or that run to the InpMaxLevels cap and stall. Most of the time the EA is doing nothing — there is exactly one Buy per fresh cycle. The interesting bars are the recovery cycles, where four positions (one initial long and up to three or four hedges) sit on the book at the same time with no protective stops, and the basket either closes at +$50 or never closes at all within the test window. If you intend to run this EA in production, the most important configuration decision is whether to set InpEnableMartingale = true and, if so, how aggressive to make the lot multiplier and the recovery step. The default configuration is the safest starting point: martingale off, single 0.10 lot, basket TP at 50. The riskier configurations are not so much a question of whether the EA 'works' as a question of whether the broker's margin and the account size can absorb a fully scaled basket that does not recover before the test ends.