Pipsgrowth EX12027 MultiIndicatorConfluence

EX12027 is built around a deliberately slow EMA pair — Fast EMA(70) and Slow EMA(290) — paired with a long-period RSI(37) that has to cross an extreme threshold (25 for buys, 70 for sells) before any signal is accepted. The combination is uncommon: the EMAs are wide enough apart to define real structural swings rather than the 5-to-50 crossovers most crossover EAs use, and the RSI thresholds are pushed to oversold/overbought rather than the more typical 30/70 midline bounces. The result is a system that holds a directional view through chop and only commits when momentum has stretched to an extremity and then begins to mean-revert toward the trend.

The signal engine has three gates. The first is the EMA trend bias, which is simply Fast EMA(70) above Slow EMA(290) for long bias and the mirror for short. The second is the RSI cross — for a buy, the previous-bar RSI value must have been below 25 and the current value at or above 25; for a sell, the previous bar must be above 70 and the current at or below 70. The third gate is an optional MACD histogram confirmation, off by default, that blocks the trade unless the histogram agrees with the direction. The MACD uses the standard 12/26/9 fast/slow/signal configuration, and you can also enforce a minimum absolute histogram value (InpMACDMinHistAbs, default 0) to require momentum of a certain size before the trade is taken. With MACD confirmation disabled, the EA trusts the EMA + RSI combination alone.

The retest (re-touch) entry filter is the EA's most distinctive optional behaviour. When InpOnlyRetestEntry is set true, a confirmed buy signal does not fire a market order immediately — instead, the EA marks the signal and waits up to InpRetestMaxBars (default 3) bars for price to pull back to within InpRetestTolerancePoints (default 150 points) of the Fast EMA(70). If the pullback arrives in time, the buy fires off the retest rather than the crossover bar; if the window expires, the pending signal is cancelled. The mirror applies to sells (a sell retest is a rally back up to the fast EMA). This is a real-world-tradable pattern: crossovers that don't get a retest tend to whipsaw, while crossovers that do get a retest tend to run.

Stops and targets have two modes. The default mode is fixed points: SL=2400 points and TP=1800 points, which yields a 1.33:1 reward-to-risk ratio against the configured SL. When InpUseATRStops is enabled, the EA switches to ATR(14) and applies a 2×ATR stop and a 3×ATR target, restoring the 1.5:1 ratio but expressed in volatility units. The ATR mode is more portable across symbols — a 2400-point stop has very different meaning on XAUUSD versus USDJPY — but the fixed-point mode is what most backtests on the default USDJPY/M5 input set have been run with. Position sizing is also dual-mode: by default the EA uses a fixed 1.0 lot per trade, but if InpUseRiskPercent is on, lots are calculated as (balance × 1%) divided by (SL points × point-value-per-lot), with the SL distance being either the ATR-derived value or the fixed 2400 points depending on the stops mode.

The dynamic lock profit (DLP) is the management layer that actually defines this EA's behaviour on winners. Once InpEnableDynamicLockProfit is on, the EA walks every open position on every tick. It measures how far price has moved in the position's favour in points, divides that move by InpLockProfitEveryXPoints (default 1500), and the integer result is the number of "lock steps" to apply. The new stop is set to entry + (steps × 1500) − 76 points, with the 76-point buffer (InpLockMinusYPointsBuffer) preventing the lock from sitting exactly on entry. The can-modify guard rejects any move that would tighten the SL by less than 0.5 points, that would put the SL inside the broker's freeze level, or that would actually loosen a buy stop or tighten a sell stop incorrectly. The net effect: every 1500 points of favourable price travel, the stop is ratcheted up by 1500 points minus the 76-point give-back. The DLP only ever ratchets in the profitable direction; the 76-point buffer keeps the trade from being closed on noise right after the lock engages.

The multi-position scaling layer permits up to 6 simultaneous trades in the same direction (MaxOpenTrades=6), but with a strict gating rule: AllowOnlyProfitableAdditions is on by default, which means a second, third, etc. position in the same direction can only be added if every existing same-direction position has moved at least MinProfitPerTradeToAdd (default 5 points) in the favourable direction. This is a real filter: it prevents the EA from averaging into a losing position during a counter-trend move. The first position in a sequence is tagged in the comment as "Initial" (or "Retest"); subsequent additions are tagged "Add#1", "Add#2", etc. The EA counts its own add-tagged positions via CountAdditionalPositions() and refuses a new add if the 6-position cap would be exceeded.

No-trade gates are layered. The spread cap (InpSpreadCapPoints, default 0) auto-resolves to 3× the rolling median spread computed over a 121-tick buffer, which is populated every tick by SpreadTrackerUpdate(). If the live spread is above that cap, no new orders are placed. The session filter is off by default but defines three discrete windows when enabled: 07:00–10:00, 13:00–17:00, and 20:00–23:00 server time, with weekends blocked regardless. Symbol trade mode is checked: if the broker reports anything other than SYMBOL_TRADE_MODE_FULL, the order is refused with a once-per-second warning. Filling mode is auto-detected per symbol via GetBestFillingMode() with IOC preferred, then RETURN, then FOK as a fallback. Slippage is capped at InpMaxSlippagePoints (default 5) and passed into the request deviation field.

The retry layer is asymmetric. Open orders go through OrderSendWithRetry with up to 3 retries, sleeping 150 ms × attempt between attempts; if the broker returns INVALID_STOPS, the EA nulls out the SL and TP and retries once more with no stops. Close retries (TryClose_EX12027) loop 3 times at 200 ms intervals, retrying on REQUOTE, TIMEOUT, PRICE_OFF, and PRICE_CHANGED. Modify retries (TryModify_EX12027) loop 3 times at 100 ms but only on REQUOTE and TIMEOUT — price-change errors are not retried, on the theory that a re-quote-on-modify with a different price should not be re-attempted blindly.

What to expect on a backtest: with the default inputs (EMA 70/290, RSI 37, 25/70 thresholds, fixed 2400/1800 stops, 1.0 lot, no MACD confirm, no retest filter, DLP on, 1500/76 lock, sessions off, spread cap auto), the EA will trade infrequently. The 70/290 EMA pair only realigns a handful of times per year on H1; on M5 the same pair whipsaws more and benefits strongly from the retest filter being enabled. The DLP layer is the dominant performance shaper — it converts winners into lock-and-hold trades rather than letting them round-trip back to the entry. A user who turns the retest filter on and disables fixed stops in favour of ATR stops gets a more portable configuration at the cost of slightly more trades per signal.