Pipsgrowth EX12004 MultiIndicatorConfluence

EX12004 MultiIndicatorConfluence is a USDJPY-tuned trend-confirmation EA built around a strict 5-of-5 AND gate. The whole strategy is reductionist: rather than scoring votes or ranking indicators, the EA waits for five independent conditions to line up on the same closed bar before it will even consider opening a position. Partial agreement — even 4 out of 5 — produces no signal. This is what makes the entry frequency low but the win-rate-versus-noise ratio disproportionately high in trending regimes on M5.

The five factors in CoreSignal() are: (1) EMA(21) versus EMA(55) direction, (2) smoothed Heikin-Ashi candle color on the closed bar, (3) RSI(14) above the bull threshold of 55.0 or below the bear threshold of 45.0, (4) MACD(12,26,9) main line above or below signal line, and (5) ADX(14) ≥ 18.0 (InpADXMin) with +DI/-DI directional agreement. Each factor contributes a single bull or bear vote, and only a clean 5-bull or 5-bear tally returns a non-zero signal. The closed bar (idx=1) is the only reference point, so signal evaluation is deterministic and not subject to intra-bar repaint.

Heikin-Ashi is constructed in-line in BuildHeikinAshi() from the last 256 M5 bars using the standard HA close/open formulas, then EMA-smoothed with smoothing period 3 (alpha = 2/(3+1) = 0.5). The result is a colour-stable filter that suppresses chop-noise on the raw candle series. Combined with the EMA cross, the Heikin-Ashi check effectively requires the price to be making higher-high/higher-low structure on a smoothed basis, not just a single-bar bounce.

Regime detection runs on every new bar through DetectRegime(), which classifies the current state into one of eight categories: STRONG_TREND (ADX ≥ 30), WEAK_TREND (ADX ≥ InpADXMin but < 30), EXPAND (ATR-percentile > 70 and Bollinger-band width expanding ≥ 1.15×), COMPRESS (ATR-percentile < 25), BREAKOUT (tight band, rising ADX), RANGE, CHOPPY, or REG_NONE. The ATR-percentile rank is computed over a 60-bar lookback (InpATRPctLookback), which means the regime classifier is adaptive to recent volatility — a market that feels calm because it was noisy last week still reads as COMPRESS in the current context.

Layer 5 — the no-trade block — is what actually gates entries. NoTradeBlock() refuses the trade when: the kill switch is on, spread exceeds InpMaxSpreadPts (35 points, USDJPY-tuned), outside the trading session (0–24 server time, so always-on but session itself doesn't actually filter), the regime is CHOPPY, the previous bar's range exceeded 3× ATR (news/big-candle guard), the cooldown timer is still active after consecutive losses, effective capital is below the InpCapitalCapFloor ($50), the realized daily loss has hit 3% of effective cap, the open-position count is at MAX_OPEN_TRADES (1), or it is Friday after 21:00 / Saturday / Sunday. The two big practical filters here are the CHOPPY regime rejection and the 3×-ATR news-bar rejection, both of which cut the failure modes that pure-confluence EAs typically bleed on.

Layer 6 — the capital cap — is opt-in. With InpCapitalCapAmount = 0.0 (default) the cap is disabled and EffectiveCapital() returns the full account equity. Set it to a positive value to cap the EA's notion of tradeable equity at that figure, which then governs both the position size (via InpRiskPercent = 0.5% of effective cap) and the daily-loss limit (3% of effective cap). The floor (InpCapitalCapFloor = 50) prevents the EA from trying to size trades on a near-empty account.

Stop and target are derived purely from ATR(14). SL = ATR × InpATRSLMult (default 2.0) floored at the broker's minimum stops distance. TP = SL × InpRRTP (default 2.0), so the structure is 1:2 R:R — a winning trade pays twice what a losing one risks, and the win-rate does not need to be 50% to be profitable. After entry, ManageExits() runs every tick and stacks four exit mechanics in order: (a) at +1R the stop is moved to break-even on the open price, (b) at +1R half the position is closed (PARTIAL_TP_PCT = 50), (c) an ATR(14) × InpATRTrailMult (default 2.5) trailing stop ratchets forward only, and (d) if the trade has been open for 120 bars (10 hours on M5) it is closed unconditionally.

The OnTradeTransaction handler walks the deal stream for DEAL_TRANSACTION_DEAL_ADD events matching this EA's magic and symbol, accumulates realized P&L per close, and triggers a 3-bar cooldown if COOLDOWN_AFTER_LOSS (3) consecutive losses occur. The cooldown is a TimeCurrent() + PeriodSeconds × 3 expiry — exactly three M5 bars in this default config. Combined with the 1-trade cap, the cooldown means a losing streak temporarily freezes the EA without flattening or reversing anything.

Two toggles in the input set deserve special attention. InpDryRun (default true) means the EA logs every entry attempt with full parameters (price, SL, TP, lots, confidence) but does not actually send the order — flip it to false to go live. InpKillSwitch (default false), when set to true, makes OnTick() call DoKillSwitch() and return before any other logic runs, which is the panic-button close-all-and-stop behaviour. The HTF confirmation timeframe (InpHTF) defaults to 8, which falls through the MapTimeframeInt switch and resolves to PERIOD_H1; the EA then requires the closed-bar M5 close to be on the correct side of the H1 EMA(50) before any entry will fire. Set InpHTF to 5 to make that explicit; lower values give the EA more signals but weaker trend context.

Order execution is the standard CTrade wrapper. SetDeviationInPoints(20) for a 20-point slippage budget, SetTypeFillingBySymbol() to auto-pick the broker's supported filling mode (FOK → IOC → RETURN), and SetMarginMode() for accurate margin checks before send. OrderCalcMargin is called explicitly before the send, and the trade is aborted if required margin exceeds free margin — a guard the typical copy-paste EA does not bother with. On entry failure the code retries once after a 150 ms sleep, then gives up. Closes and modifies are wrapped in 3-retry loops with 200 ms / 100 ms sleeps on REQUOTE, TIMEOUT, PRICE_OFF, and PRICE_CHANGED.

The OnTester custom criterion is (net × profit factor) / (1 + |balance drawdown|), with a 30-trade minimum (trades below that return 0). This formula rewards both raw profit and consistency, and penalizes drawdown linearly — meaning a strategy with 2× the net profit at 3× the drawdown will score lower than a smoother strategy with half the profit. For a 5-of-5 confluence EA on M5 USDJPY, expect trade frequency in the 2–8 trades-per-day range, concentrated in the London and New York sessions, with a strong dependency on regime — most of the equity curve is built during the ADX > 25 windows, and the choppy, ranging weeks are mostly dead capital preserved by the no-trade block.