Pipsgrowth EX12002 MultiIndicatorConfluence

EX12002 is a vote-counting confluence engine, not a single-signal EA. It stacks eight independent technical reads on every closed bar, lets each one cast a +1 or -1 vote, and only fires a trade when at least five of those eight votes point the same way. The eight voters are: the 21/50 EMA cross (fast above slow = bullish vote), the slope of a 20-period Hull moving average, the 2/30/10 Adaptive Moving Average's position relative to the prior bar's close, a Heikin-Ashi close-above-open check, a 10-period/3.0x ATR SuperTrend band test on the bar's midpoint, RSI(14) above or below 50, MACD(12,26,9) main line vs signal, and an Adaptive Hull computed as 2×LWMA(N/2) − LWMA(N) with slope direction. Each voter can add a +1 or a −1 (long or short) vote. The magnitude of the vote sum is converted to a confidence score with confidence = min(100, |votes| × 12), so an 8-of-8 sweep scores 96, a 5-of-8 minimum scores 60, and a 4-of-8 disagreement scores 0 and the trade is blocked. The threshold is exposed to the user as InpMinConfluenceVotes (default 5) so traders can dial the bar up to 7 or 8 to filter for the highest-conviction bars only.

Before any signal vote is counted, the EA classifies the current market regime from the entry-timeframe indicators. DetectRegime() pulls the 14-period ADX, the 14-period ATR, and a 20-period/2.0σ Bollinger band; from those it builds a 100-bar ATR percentile (compressed if below 25, expanded if above 85) and a 100-bar BB-width percentile, then maps them to one of seven states: REG_STRONG_TREND (ADX≥25 + ATR expanded), REG_WEAK_TREND (ADX≥20 + ATR normal), REG_COMPRESS (ADX<18 + ATR compressed), REG_EXPAND (BB-width expanded + ADX<20), REG_BREAKOUT (BB-width expanded + ADX≥20), REG_RANGE (ADX<18 fallback), and REG_CHOPPY. The CHOPPY and NONE states are hard-blocked by RegimeAllowsEntry() — even an 8-of-8 vote will not override a CHOPPY regime call. This is a deliberate guard: the EA is built for trends and clean two-sided ranges, not the noise that follows a stop-run or news event.

Entry is gated by a second function, ConfirmEntry(), which is independent of ComputeSignal(). The confirm function checks four things in order: (1) higher-timeframe EMA agreement — if InpRequireHTFAgreement is true, the M30 EMA(21) must be above the M30 EMA(50) for a long, below for a short; (2) candle-close direction — if InpUseCandleConfirm is true, the prior bar's close must be above its open for a long, below for a short; (3) spread — the current ask-bid spread must be below 3.0×30 points, an unusually wide spread is rejected; (4) session — server hour must be between 7 and 20. The HTF check is what makes the EA's entries look 'patient' in backtests: a 7-vote confluence on the M1 candle is still skipped if the M30 trend is the wrong way. Traders who want the EA to fire more often can flip InpRequireHTFAgreement to false, but the default is true.

A third gate, NoTradeReason(), runs after the confirm passes. It is the consolidated safety stack: capital-cap floor check (blocks if effective capital is below the $50 floor), capital-cap ceiling check (treats equity as min(cap, equity) when InpCapitalCapAmount > 0), session guard, spread guard, cooldown-after-losses guard (set by InpCooldownAfterLoss consecutive losing closes from the day's deal history), daily realized P&L guard (3% of effective capital, computed live from the deal stream in RefreshTodayRealized()), weekly realized P&L guard (6%, computed by ComputeWeeklyRealized() which walks back to Monday's midnight), portfolio heat guard (InpMaxOpenTrades=5), and per-symbol guard (InpMaxSymbolTrades=3). If any of these return a non-empty reason string, the entry is blocked and the reason is printed to the log so the trader can audit why a bar was skipped.

Position sizing is risk-percent based, not fixed-lot. CalcLotByRisk() takes the effective capital, the stop distance in price, the symbol's tick value, and the symbol's tick size, and returns riskMoney / ((slDist / tickSize) × tickValue) — i.e. the number of lots that lose exactly InpRiskPercent of effective capital if the stop is hit. InpRiskPercent defaults to 0.5%, which is conservative for an M1 XAUUSD confluence EA. The lot is then clamped to the broker's min/max via NormVol(), which floors to the symbol's LotsStep and snaps to the broker's LotsMin/LotsMax. A MarginOK() pre-send check rejects entries that would consume more than 80% of free margin, preventing a margin call from a wide-spread or gap entry.

The SL is ATR-based, not point-based. TryEntry() reads the ATR(14) on the prior bar, multiplies by ATR_SL_MULT=2.0 to get the stop distance, and floors at StopsLevelPrice() + 1×point so it never sits inside the broker's minimum stop band. The TP is ATR_TP_MULT=3.0 × ATR(14) — that is a 1:1.5 risk-to-reward ratio by construction, and it is the most consequential single parameter in the EA's behavior. Traders who want a wider TP can edit ATR_TP_MULT directly in the source (it is a #define, not an input), or flip InpUseATRSL to false to fall back on a 100-point fixed stop on the entry timeframe. The latter is rarely useful on M1 XAUUSD; the ATR path is the intended one.

Once a position is open, ManageOpenPositions() runs on every tick and applies four layers of management. Layer 1: regime-change exit — if DetectRegime() returns CHOPPY or NONE, the position is closed with the reason 'regime change'. Layer 2: opposite-signal RSI exit — a long is closed if RSI(14) drops below 30, a short is closed if RSI(14) rises above 70. Layer 3: time-based exit at 500 bars on the entry timeframe (M1 = 8 hours 20 minutes, M5 = ~41 hours). Layer 4: one-shot break-even at 1.0R — when price moves one full initial-risk in the trade's favor, the SL is moved to entry + 5 points and never moves back. The break-even is genuinely one-shot: it triggers once per ticket and does not ratchet forward as price keeps moving.

After the BE has fired, partial-take-profit at 50% of the trade's TP distance kicks in if the remaining position volume is at least 2× the symbol's minimum lot. PARTIAL_TP_FRACTION=0.5 means half the position is closed at the halfway point to TP. After the partial, an ATR trailing stop takes over: the new SL is cur ± ATR(14) × 3.0 and is ratcheted forward only (a long's SL can move up, never down). The 3.0×ATR trail is deliberately wide — a 1.5× or 2.0× trail would be too tight for an M1 XAUUSD EA, which routinely sees 50–80 point M1 candles. The 3.0× trail gives trades room to breathe through pullbacks.

Pyramid / scaling is present in the code (TryPyramid()) but disabled by default (InpPyramidEnabled=false). When enabled, it adds one position per tick when (a) the regime is not CHOPPY, RANGE, or COMPRESS, (b) the symbol already has fewer than PYRAMID_MAX_LEVELS=3 positions, (c) the existing position is at least 1.5×ATR(14) in profit, and (d) margin is OK. The add-on lot is the existing position's volume (multiplier 1.0), so it is a true scale-in, not a martingale. A capital-cap opt-in (InpCapitalCapAmount=0 means off, any positive value caps effective capital at that dollar amount) lets a trader with a $10k account pretend their account is $2k for sizing purposes, which is useful for prop-firm accounts or for testing a small-account risk profile on a larger balance.

InpDryRun defaults to true — the EA will compute entries, exits, and management logic on every tick and print them to the log, but will not send any order. To go live, the trader must explicitly flip InpDryRun to false in the inputs. This is a deliberate safety default. The companion InpKillSwitch does the opposite: when flipped to true, it forces-close every open position for this magic number on the next tick and blocks any new entries. It is intended as a panic button — wire it to a chart button or external alert and you have a one-click 'get out'.

The OnTester formula is (netProfit × profitFactor) / (1 + balanceDrawdownDollars), with a hard floor of 30 trades (trades < 30 → 0). This is a stricter fitness function than most EAs in the family — the 30-trade minimum means single-bar scalps that happen to net-positive will be ranked below steady 100-trade runners, and the 1 + DD denominator (not 1 + DD%) means drawdown in absolute dollars is what hurts the score. Optimizer passes that net $1,000 with a $5,000 DD rank higher than passes that net $800 with a $200 DD. The implication for live use: the optimizer will favor robustness over peak profit, which is the right trade-off for an 8-vote confluence system on M1 XAUUSD.