Pipsgrowth EX18105 TrendFollow

Pipsgrowth EX18105 TrendFollow is an fxDreema-built Expert Advisor that turns a textbook SMA crossover into a 9-leg stacked entry, then manages the resulting basket with a server-agnostic virtual trailing stop and a 7-day order expiry. The filename suffix MM3_0_FLIP_CODEPRO is the visual-block template that generated the .mq5 — every trading decision runs through a graph of 46 CBlockCalls derivatives (Block0 through Block45) wired by the fxDreema export rather than hand-written MQL5 logic. The visible behavior, however, is straightforward: a single bar's SMA(10) vs SMA(50) relationship decides whether the EA buys, sells, sits flat, or flattens an open basket.

Entry is gated by Block0, a no-position check that iterates TradesTotal() and reports whether anything is open on the chart's symbol. When Block0 reports no position, control passes to Block2 (long condition) and Block3 (short condition). Block2 is an MDL_Condition wrapper around two MDLIC_indicators_iMA instances: Lo runs iMA(_Symbol, _Period, 10, 0, MODE_SMA, PRICE_CLOSE, 1) and Ro runs the same call with MAperiod=50; both use Shift=1, so the comparison is between the SMA(10) and SMA(50) of the prior closed bar. Block2's compare operator is ">" and Block3's is "<" — the relationship inverts cleanly. A long crossover fans out to nine parallel buy blocks (Block4, Block10, Block14, Block18, Block22, Block24, Block26, Block28, Block30, Block32) and a short crossover fans out to nine mirror sell blocks (Block5, Block11, Block15, Block19, Block23, Block25, Block27, Block29, Block31, Block33). Each parallel leg is its own MDL_BuyNow or MDL_SellNow instance with a distinct volume extern — inp11_VolumeSize through inp118_VolumeSize — all defaulted to 0.01 lots in the input panel. MagicStart=22218105 is added to the leg's Group integer, so the broker sees nine separate position tickets per signal but they all belong to this EA's namespace.

Exit logic is symmetric. Block1, the if-position check, runs in parallel with Block0 and cascades to Block8 and Block9 — the same SMA(10) vs SMA(50) condition classes with the inverted comparison. When SMA(10) crosses below SMA(50) on the closed bar, Block8 fires Block6, an MDL_CloseOpened instance that walks the position list and closes every long ticket belonging to the magic namespace. Block9 handles the opposite direction with Block7 closing shorts. There is no fixed take-profit — TakeProfitMode is set to "none" in every buy and sell block, and StopLossMode is also "none" on the live ticket. The only protective layer is VirtualStopsDriver, which is invoked on every OnTick() when USE_VIRTUAL_STOPS is true. VirtualStopsDriver draws a trendline object on the chart at trade entry and uses it as a virtual stop: once price moves 50 pips in profit (inpTrailStartPips=50.0), the trailing line ratchets 10 pips (inpTrailStepPips=10.0) and parks 20 pips behind price (inpTrailStopPips=20.0). Because the stop is virtual, the broker never sees a server-side SL — a chart crash or terminal disconnect erases the trailing line.

Order expiration is built into the MDL_BuyNow and MDL_SellNow templates: dExp.ModeTimeShift=2 (relative shift), dExp.TimeShiftDays=1.0, TimeSkipWeekdays=true, then ExpMode="specified" with ExpDays=7, ExpHours=0 on top. The effective behavior is "expire 7 days from entry, skipping weekends." Position-level modifications pass through TryClose_EX18105, TryClosePartial_EX18105, and TryModify_EX18105 retry helpers, each running up to three attempts at 100–200 ms intervals to absorb transient requotes.

A few quirks are worth knowing. ENABLE_SPREAD_METER is true, so the EA draws a live spread overlay on the chart during non-tester runs; ENABLE_STATUS draws a top-left status banner. OnTester() is not implemented, so optimization is driven by the strategy tester's standard profit/drawdown objective. The brief in the source header claims an "AI SAR" entry, but no Parabolic SAR call exists in the code — the actual signal is the iMA SMA(10)/SMA(50) crossover, so trust the source rather than the marketing line. Because every position exits only when the SMA flips or the 7-day timer fires, the EA's holding period is structurally long and is a poor match for traders who expect minute-scale scalping turnover despite the "scalper" label in the brief.

The nine-parallel-leg design is the defining structural choice. In a single buy event the EA opens 9 × 0.01 lots = 0.09 lots of exposure with the same magic prefix, all tracking the same SMA crossover. The virtual trailing stop treats them as one combined position (VirtualStopsDriver walks all matching tickets), and the close block flattens all nine in a single pass. Operators who want tighter risk should lower all nine inp_VolumeSize externs together — the structure is monolithic, so changing one leg's size without the others creates an unbalanced basket.

For backtesting, the lack of an OnTester() function means MT5 falls back to its built-in optimization objective, which by default weights profit factor and total net profit. Because all nine legs open simultaneously on every crossover, the resulting equity curve tends to step up on each clean SMA(10)/SMA(50) flip and bleed slowly between signals; this is the visual signature of a basket system rather than a single-ticket EA. The spread overlay (ENABLE_SPREAD_METER) and the top-left status banner (ENABLE_STATUS) are only drawn outside the strategy tester, so visual sanity checks require a demo account run. ENABLE_TEST_INDICATORS controls whether tester-mode charts still render the FXD graphical objects, but the only indicator the EA actually creates is the iMA buffer inside the MDL_Condition wrappers — there are no separate iCustom handles, no chart-attached oscillators, and no subwindow panes to inspect.

The OCO (one-cancels-other) driver runs in the background of OnTick() and is the silent backstop for accidental duplicate positions: if a tick fires Block0's no-position check while a residual position is being closed, OCODriver() will keep the open and the close request from racing. The OCO behaviour is automatic and not exposed via inputs, so it is worth knowing it is there but not worth tuning.