Pipsgrowth EX12053 MultiIndicatorConfluence

Pipsgrowth EX12053 is a lean Heikin-Ashi-led trend-catcher for MetaTrader 5, built around a deliberate choice: keep the indicator stack small, keep the rules readable, let price action do most of the work. Where most MultiIndicatorConfluence variants in the same family pull in six to ten native indicator handles, EX12053 stands up exactly three — an EMA(21) for short-term trend, an EMA(50) for the structural backstop, and a MACD(12,26,9) for momentum confirmation — and pairs them with Heikin-Ashi direction computed inline from raw price data. The whole trade thesis is: only take a position when smoothed-price direction (HA), trend-structure direction (EMA cross), and momentum direction (MACD main vs signal, with a sign-bounded filter) all agree, and the fast EMA has actually been climbing or falling for at least the last few closed bars.

The signal chain lives in GenerateSignal(). On each new bar the EA pulls the last two closed-bar values of every handle via ReadIndicatorsOnClosedBar(), calls GetHeikinAshiOnBar(1, ...) to compute Heikin-Ashi on the most recent closed bar from a 200-bar CopyRates window, and runs the four sub-conditions for a long entry in this order: HA must be bullish (HA close above HA open, where HA close is the average of the four OHLC prices and HA open is the midpoint of the previous HA bar), the fast EMA(21) must print above the slow EMA(50), the MACD main must be above its signal line AND above zero (so it's not just a bullish cross inside a bearish regime — it has to be a positive-side bullish cross), and EMASlopeOK(true) must return true, meaning at least MinEMASlopeBars-1 of the last MinEMASlopeBars slope checks show the fast EMA rising. The short side is the perfect mirror: HA bearish, fast EMA below slow, MACD main below signal AND below zero, fast EMA falling on the slope check. If even one of the four sub-conditions fails, no signal fires. This is the design choice that defines EX12053 — it will not enter on a "decent" trend, only on a trend that every smoothed layer agrees on at the same instant.

Heikin-Ashi is computed from scratch on every signal bar in GetHeikinAshiOnBar(). The function allocates dynamic arrays sized to a 200-bar copy of MqlRates, seeds the first HA bar with the real open and the OHLC/4 average, then iterates forward using the canonical recursive formulas (HA close = OHLC/4; HA open = (prev HA open + prev HA close) / 2). The EA does not use an external Heikin-Ashi indicator — it does the math itself, which means the calculation is fully transparent and the only cost is the 200-bar rate copy. The HA bar is then compared to its own open to derive a directional bias (haBull / haBear), and that single boolean is the first gate in the signal chain. In a sideways market where raw candles alternate red and green, smoothed HA tends to flatten or flip back and forth, and the first gate alone blocks most of the chop.

The MACD sign-bounded filter (macdUp = (mMain > mSig) && (mMain > 0.0)) is the second defining choice. A plain MACD cross can fire anywhere on the indicator, including inside a deep bearish pullback of an otherwise bullish trend, which historically produces a lot of false signals on gold. By requiring the MACD main to also be on the correct side of zero, the EA restricts entries to cross events that occur in the "right half" of the histogram — the half where momentum is actually supporting the trend, not just turning inside a dead zone. Combined with the EMA cross and the slope check, this means the MACD line has to be moving in the trade direction, on the correct side of zero, while the fast EMA is sloping in the trade direction, while HA is bullish or bearish. Three independent momentum/trend layers stacked in series is a much higher bar than any one of them alone.

Entry execution goes through PlaceEntry(), which is gated by PreflightAllowsEntry(). The pre-flight check evaluates four conditions in order: the EA must not be stopped (terminal disconnect, etc.), InSession(TimeCurrent()) must return true, SpreadOkay() must return true, and the current time must be past g_tradePauseUntil (set by the loss-driven pause guards). The session filter uses a single server-time window with SessionStartHour=7 and SessionEndHour=22, and the helper InSession() handles the midnight-wrap case automatically if the window ever spans 24 hours. The spread gate is a median-based cap: PushSpreadSample() keeps a rolling window of the last SpreadMedianWindow=20 ticks of spread, MedianSpread() sorts and returns the median, and the current spread must be at or below SpreadCapMultiplier=3.0 times that median — i.e., the current spread can be up to 3× the recent typical spread before entries are blocked. This is a permissive filter compared to fixed-point caps; it lets the EA trade through normal news-widening events but rejects pathological spikes. If pre-flight clears, the EA computes the SL and TP from StopLossPoints=300 and TakeProfitPoints=600 and sends the order through SendMarketOrder().

The order sender implements a three-pass retry loop, with each pass cycling through the three filling modes ORDER_FILLING_FOK → ORDER_FILLING_IOC → ORDER_FILLING_RETURN. If the broker rejects the order with TRADE_RETCODE_INVALID_FILL or TRADE_RETCODE_INVALID, the EA steps immediately to the next filling mode (no sleep); if it gets a requote or price-off, it refreshes the price and retries; between full passes it sleeps RetryDelayMs=250 ms. After a successful fill, the EA scans open positions for the matching magic and ensures the inline SL/TP were actually applied (some ECN brokers ignore inline stops); if they weren't, it sends a TRADE_ACTION_SLTP modify. This is a small but important detail for XAUUSD on ECN accounts, where SL/TP drops are a common cause of unexpected losses.

Position management is split into two pieces. First, the per-trade DLP step-ratchet (ApplyDynamicLockProfitToPosition()) runs on every tick against every owned position. It computes movedPts from entry to current bid/ask, divides by InpLockProfitEveryXPoints=150 and floors it to get the number of completed lock steps, sets lockFromEntryPts = steps × 150 - InpLockMinusYPointsBuffer=20, and proposes a new SL at entry ± lockFromEntryPts. The proposed SL is then clamped against SYMBOL_TRADE_STOPS_LEVEL and refused if it would land inside SYMBOL_TRADE_FREEZE_LEVEL. Critically, the ratchet only ever tightens — if(curSL > 0 && newSL <= curSL + _Point*0.5) return; — so a sudden spike can never loosen the stop. After 150 points of progress, the SL moves to entry+130, after 300 points to entry+280, after 450 to entry+430, and so on. Second, the scaling gate in PlaceEntry() enforces the Initial vs Add#N hierarchy: the first same-direction trade gets the comment tag Psgrowth.com Expert_12053 Initial BUY/SELL; every additional same-direction trade is gated by IsAllSameDirProfitableAtLeast() which checks that every existing same-direction position is at least MinProfitPerTradeToAdd=5.0 points in profit, then gets the tag ... Add#N BUY/SELL with g_addCountBuy/g_addCountSell as the per-direction counter. The AllowOnlyProfitableAdditions flag (default true) controls whether the gate is enforced. Total same-direction scaling is capped at MaxOpenTrades=5.

The pause-guard system runs from OnTradeTransaction() on every deal-add event. When a deal entry of type DEAL_ENTRY_OUT (i.e. a closing deal) lands and the deal belongs to this EA's magic and symbol, the EA reads DEAL_PROFIT + DEAL_SWAP + DEAL_COMMISSION for the round-turn PnL. If the PnL is negative, g_lossStreakCount increments and the loss amount is added to g_lossStreakAmt; if it's positive or zero, both reset to zero (one win wipes the streak). Two independent trigger paths then check the running streak: PauseOnConsecutiveLosses (default OFF) at PauseConsecutiveLossesCount=3 losses sets g_tradePauseUntil = TimeCurrent() + PauseConsecutiveLossesMinutes*60; PauseOnLossAmount (default OFF) at PauseLossAmount=100 USD sets the same kind of pause for PauseLossAmountMinutes=60 minutes. The two paths stack via MathMax, so triggering both gives the longer pause. Pause logging is throttled to once every 15 seconds to keep the Experts tab clean.

On the risk and sizing side, EX12053 offers two modes. The default is UseRiskPercent=false with FixedLot=0.10, which on a $100 account with a 300-point SL on XAUUSD represents roughly 30% margin on a typical gold contract — comfortable for a single-position lot, but with the 5-trade scale-in cap the EA can stack up to 0.50 lots. The second mode is UseRiskPercent=true with RiskPercent=1.0, in which case LotsByRisk() computes the lot size from AccountInfoDouble(ACCOUNT_BALANCE) * (RiskPercent/100) / (sl_points * valuePerPointPerLot), snapped to the symbol's SYMBOL_VOLUME_STEP and clamped to SYMBOL_VOLUME_MIN / SYMBOL_VOLUME_MAX. The function ClampVolume() does the snap-and-clamp math. There is no ATR-sizing, no per-trade risk override, no equity curve filter — the EA is intentionally small. The CTrade wrappers TryClose_EX12053, TryClosePartial_EX12053, and TryModify_EX12053 are present for any external utility hooks, each with a 3-attempt loop and 200ms/100ms sleep on requote or timeout.

The combination of a 14-input control surface, a 3-handle indicator stack, and a 4-condition signal makes EX12053 the most approachable member of the MultiIndicatorConfluence family. There is nothing in this EA that an intermediate trader cannot read end-to-end: the inputs are exactly what they say, the signal logic is the same code in GenerateSignal() you would write on a whiteboard, and the risk rules are explicit. What the EA gives up by being small is breadth — there is no regime classifier, no on-tester optimization, no pending-order infrastructure, no AutoTune volatility adaptation, no separate per-direction EMA — and what it gains is transparency. The HA-led 4-condition confluence is the design choice that ties it together: a single, testable, repeatable rule about what counts as a valid trend entry, applied on every new bar.