Pipsgrowth EX12029 MultiIndicatorConfluence

Pipsgrowth EX12029 is a thin, two-signal trend-following Expert Advisor built around one crossover pair and one oscillator threshold. The whole decision engine in OnTick is fed by three indicator handles created in OnInit: a 70-period EMA (hEMA_Fast), a 290-period EMA (hEMA_Slow), and a 37-period RSI (hRSI) — all on close. There is no MACD, no ATR stop mode, no retest, no HTF filter, and no regime classifier; the 12-layer breakdown in the brief is the architectural template the family follows, not a list of switches wired into this specific build. What you actually get is the entry logic: SignalBuy() requires RSI[1] < 25 && RSI[0] >= 25 while TrendBiasBuy() (EMA70 > EMA290) is true; SignalSell() mirrors it with the 70-threshold and a fast-below-slow EMA read. Signals are evaluated only on a new bar (IsNewBar()) so a sustained move that doesn't print a fresh RSI threshold cross on the closing bar of a new candle simply doesn't trade.

Position management runs the same way: minimal and explicit. The first order in a direction is tagged "Initial BUY" or "Initial SELL"; every subsequent same-direction order is tagged "Add#1", "Add#2", and so on. Before any add is allowed, IsAllSameDirProfitableAtLeast(dir, MinProfitPerTradeToAdd=5.0, tick) walks the open ticket pool and requires every existing same-direction position to be at least 5 points in the green — so the EA will not stack a losing position on top of a losing position. The hard cap is MaxOpenTrades=6 per magic number, counted across both directions by CountAllOpenPositions(). This makes the strategy a low-trade-frequency trend rider, not a scalper: it places one new entry per qualifying bar until it hits the cap, then waits.

The dynamic-lock-profit layer is the only thing doing work between entries. Every tick, the OnTick maintenance loop calls ApplyDynamicLockProfitToPosition() for each open ticket. That function measures how far price has moved from entry in points (moved), divides by InpLockProfitEveryXPoints=1500, and if the integer number of "lock steps" is at least one, it sets a new stop at entry + steps*1500 - 76 for a buy (or the mirror for a sell). 76 points is the give-back buffer that prevents the stop from sitting on the high-water tick. The function is strictly one-way: CanModifySL() rejects any new SL that isn't tighter than the current one (within a 0.5-point noise band), and the new SL is also clipped to respect the broker's SYMBOL_TRADE_STOPS_LEVEL and SYMBOL_TRADE_FREEZE_LEVEL so the modify doesn't get bounced. After the floor, ModifySLTP() retries up to three times on REQUOTE or TIMEOUT, sleeping 120ms×attempt between tries. Net effect: on USDJPY the SL ratchets at every 150-pip leg of profit, and it will never loosen again for the life of the trade.

The risk inputs are deliberately portable. SL is InpSLPoints=2400 and TP is InpTPPoints=1800, which gives a 0.75:1 R:R on a fixed SL/TP basis. With the 1.0-lot default and InpUseRiskPercent=false, sizing is constant; flip the switch and CalcLotFromRisk() derives lots from balance * InpRiskPercent / (SL_points * valuePerPointPerLot) and normalizes to the symbol's SYMBOL_VOLUME_STEP. The default InpMaxSlippagePoints=5 is passed straight into req.deviation on every OrderSend. OrderSendWithRetry() itself retries three times on the generic retcode path and will also, on a TRADE_RETCODE_INVALID_STOPS, strip the SL/TP and re-send once before giving up. Filling mode is chosen by GetBestFillingMode() in this order: IOC → RETURN → FOK → RETURN fallback, so it adapts to whatever the broker allows.

The session filter is a free feature sitting behind a disabled switch. InpUseSessionFilter=false by default, which means MarketIsTradableNow() short-circuits to true and trades 24/5 minus the broker's own tradable window. When you flip it on, CacheSessionConfiguration() parses three "HH:MM" strings into minutes-from-midnight and IsWithinTradingSessions() runs a 3-window OR check that also handles the wrap-around case where start > end (an evening→morning window). The three defaults — 07:00–10:00 (London morning), 13:00–17:00 (London/NY overlap), 20:00–23:00 (Asian evening) — are pre-loaded but inert until the user enables the filter, and weekends are blocked by day_of_week == 0 || 6 regardless of the windows. The spread gate is wired in directly: CheckSpreadOK() blocks entries whenever (ask-bid)/point > InpMaxSpreadPoints=20, so XAUUSD during a wide-spread NY rollover can sit on its hands even with the session filter off.

Ticket resolution is the one place this EA does extra work that the rest of the family sometimes skips. After OrderSendWithRetry() returns a MqlTradeResult, OpenPosition() tries HistoryDealSelect(res.deal) to read the DEAL_POSITION_ID, then HistoryOrderSelect(res.order) to read ORDER_POSITION_ID, and then waits up to 5×40ms for PositionSelectByTicket() to confirm the position is visible. If that still fails it falls back to a PositionsTotal() walk matching on symbol, magic, comment, and lot. This is the part that makes the EA robust against brokers that take a few ticks to make a new position visible to PositionSelectByTicket() — which matters because ApplyDynamicLockProfitToPosition() runs on the next tick, not the same tick, and a missing ticket would silently skip the ratchet.

The execution tail is short but explicit. TryClose_EX12029() and TryClosePartial_EX12029() both retry three times on REQUOTE / TIMEOUT / PRICE_OFF / PRICE_CHANGED with a 200ms sleep, while TryModify_EX12029() retries three times on REQUOTE / TIMEOUT only with a 100ms sleep — modify is the asymmetric one because a PRICE_CHANGED on a stop-loss modify is treated as a broker rejection, not a transient. The bottom line: EX12029 is the kind of EA you point at one symbol on one timeframe, set a sensible max-trades cap, leave DLP on, and let it ride. The reason it's tagged "slim" in the family is exactly that — no auxiliary filters to debug, no regime state machine to validate, just an EMA cross with an RSI gate and a 1500/76 ratchet.