Pipsgrowth EX12077 MultiIndicatorConfluence

Pipsgrowth EX12077 Pattern Classifier is a deliberately narrow tool. It watches XAUUSD on the M5 timeframe for one specific event — a bullish or bearish engulfing candle that closes on the correct side of a 50-period EMA — and fires a single trade in that direction the moment the pattern prints. There is no grid, no martingale, no multi-symbol portfolio, no pyramid, no recovery sequence. The whole decision tree lives in seven functions and reads from a single iMA handle, which keeps the EA small enough to audit line by line and makes it a useful baseline against which the more elaborate multi-confluence EAs in the same family can be measured.

The signal logic sits in CheckSignal(). On every new bar, the function reads the open, high, low, and close of bar 1 (the just-closed bar) and bar 2 (the bar before it), pulls one buffer from the EMA handle into bufMA, and applies two boolean tests. A bullish setup requires that bar 2 is red (close below open), bar 1 is green (close above open), bar 1's close exceeds bar 2's open (body engulfs body), and bar 1's close is above the EMA — a "simplified engulf" as the source comment notes, since the high/low don't have to be engulfed, only the bodies. A bearish setup mirrors that: bar 2 green, bar 1 red, bar 1's close below bar 2's open, and bar 1's close below the EMA. The InpUseEngulfing input (default true) gates the whole pattern — flip it false and CheckSignal() returns without setting either flag.

Three gates sit between the signal and the order. The first is a spread cap at 250 points (InpMaxSpread), explicitly flagged in the input comment as "Increased for Gold" — XAUUSD routinely widens to 25-40 points during New York opens and London fixes, so a tight cap would block most sessions; 250 points (≈25 pips on a 5-digit gold quote) is a pragmatic ceiling that keeps the worst fills out without gating the day. The second is the new-bar guard inside OnTick() — static datetime lastBar paired with iTime(_Symbol, _Period, 0) — so CheckSignal() runs at most once per candle, not on every tick. The third is if(PositionsTotal() > 0) return at the top of CheckSignal(): the EA only ever holds one position at a time. There is no pyramiding into winners, no scaling into losers, and no second-entry logic.

Risk management runs through CalculateLotSize(). With InpRiskPercent = 1.0 (default), the function reads account balance, divides by 100 to get 1% of equity, then converts the 400-point stop into a per-lot dollar risk using the symbol's tick value, tick size, and point. The result is floored to the symbol's lot step with MathFloor(lot/step)*step, then clamped between the broker's LotsMin and LotsMax. Set InpRiskPercent = 0 and the function falls back to InpFixedLot = 0.01. Either path gives a deterministic, broker-aware lot — no arbitrary multipliers, no recovery ramps.

The trade itself is a clean 1:2 risk-reward. OpenBuy() computes stop loss as Ask − 400 * Point and take profit as Ask + 800 * Point. OpenSell() mirrors with bid-based math. The trade.Buy(...) and trade.Sell(...) calls pass the comment string "Psgrowth.com Expert_12077" so fills are easy to filter in journal reports, and magic 22212077 keeps them in their own bucket apart from any other EA running on the same account. The CTrade object is configured at OnInit with deviation 3 points and ORDER_FILLING_FOK — fill-or-kill means a partial fill rejects the whole order, which avoids the half-filled position state that some brokers can produce on thin gold books.

The retry helpers TryClose_EX12077, TryClosePartial_EX12077, and TryModify_EX12077 are defined at the bottom of the file. Each loops up to three times on REQUOTE, TIMEOUT, PRICE_OFF, and PRICE_CHANGED retcodes with a 200 ms backoff (100 ms for the modify variant). They are not wired into the live OpenBuy / OpenSell / CheckSignal paths in this build — the EA's actual close and modify calls go straight through trade.PositionClose and trade.PositionModify. The helpers remain available for a one-line patch if the user wants to harden the exit path against requotes during volatile gold sessions.

Indicator stack: a single iMA(_Symbol, _Period, InpTrendEMA=50, 0, MODE_EMA, PRICE_CLOSE) handle initialised in OnInit and released in OnDeinit via IndicatorRelease(hMA). The handle pulls one buffer (index 0) on demand into bufMA with CopyBuffer(hMA, 0, 1, 1, bufMA), reading bar 1's value at the moment of the new bar. The whole trend filter is a single exponential moving average — no ATR, no ADX, no RSI, no volatility gate. The classification is binary: either the engulfing pattern is on the right side of the EMA or it is not.

Backtesting posture: the source header advertises the EA as part of a "12 layers" stack (REGIME, SIGNAL, ENTRY, CONFIRM, NO-TRADE, CAPITAL CAP, RISK, SIZING, MANAGE, EXIT, SCALING, OnTester) but only a subset of those layers is actually implemented in this build. There is no custom OnTester() function, so the strategy tester falls back to the standard balance / max-DD / profit-factor / Sharpe scoring. There is no per-day or per-week loss cap, no equity-stop, no global drawdown circuit breaker. The hard caps are the 250-point spread gate, the one-position-at-a-time limit, and the 400-point per-trade stop. That makes the EA transparent — what you see in the source is what runs in the tester — but it also means the risk ceiling is the broker's margin call, not an internal protection.

What to expect in a forward test: trade frequency will be low. Gold M5 only prints a body-engulfing bar on the correct side of a 50-EMA on a fraction of all bars, and the one-position-at-a-time cap means the EA waits for a close before re-entering. Expect the equity curve to look like a string of small 1:2 wins interrupted by occasional full-stop losses, with long flat stretches during Asian-session consolidation when the small-bodied bars do not satisfy the engulf condition. The pattern is more useful as a session filter for a discretionary trader — a way to align with a momentum bar in the direction of a slow EMA — than as a stand-alone income stream, but the engineering is honest: no hidden lot multiplier, no martingale, no curve-fitted add-on, no equity magic.