Pipsgrowth EX09023 MeanReversion

Pipsgrowth EX09023 BollingerMeanReversion fades outer Bollinger Band touches on a single-instrument basis. The strategy is built around a deliberately small signal stack: a single iBands(20, 2.0, PRICE_CLOSE) handle, no confirmation indicator, and no regime classifier. When the bid reaches or pierces the lower band, the EA buys with the BB middle as the profit target. When the bid reaches or pierces the upper band, the EA sells with the BB middle as the profit target. SL is fixed at 30 pips; lot is fixed at 0.10; one position at a time. The whole design is essentially a minimum-viable mean-reversion loop, and the article below walks through every actual code path so traders can verify the claim before going live.

The header in the source comment advertises a 12-layer stack: REGIME, SIGNAL, ENTRY, CONFIRM, NO-TRADE, CAPITAL CAP, RISK, SIZING, MANAGE, EXIT, SCALING, OnTester. Of those twelve, only seven have actual logic. SIGNAL and ENTRY (the BB-touch trigger), NO-TRADE (an 8-gate IsSafeToTrade), CAPITAL CAP (InpCapAmount / InpCapFloor), RISK (3% daily loss limit, 5.0 max lot), SIZING (fixed lot 0.10), EXIT (the SL plus a dynamic BB-middle TP), and OnTester are implemented. REGIME (no classifier), CONFIRM (no second-stage filter), MANAGE (no trailing, no break-even, no partial close), and SCALING (no pyramid, no martingale, no grid) are missing. That is not an oversight — the EA's stance is that any extra layer would dilute a clean fade, and the header is intentionally left as a marketing-style aspirational list rather than a contract.

The entry logic lives in OnTick. The tick handler runs UpdateDailyCounters_EX09023 (rolls g_tradesToday, g_consecLosses, g_realizedToday to zero on a new calendar day), then checks HasTrade() to enforce single-position control. HasTrade scans PositionsTotal from the top, looking for any open position with magic 22209023 on the chart symbol; if one is found, the tick returns immediately. The current bar's open time is read; if it matches the lastBar timestamp, the tick returns — signal evaluation only runs on bar close, not on every tick. The 8-gate IsSafeToTrade_EX09023 must clear, then the EA copies buffer 1 (upper), buffer 2 (lower), and buffer 0 (middle) from the BB handle for bar 1, the just-closed bar. If bid <= lower[0], a buy loop runs three times: each iteration recomputes OrderCalcMargin for ORDER_TYPE_BUY at the current ASK, checks ACCOUNT_MARGIN_FREE, and calls TryBuy_EX09023; between iterations the loop sleeps 500ms. The mirror logic fires on bid >= upper[0] via TrySell_EX09023. Both Try functions accept the BB middle as the TP argument.

A subtle but important design choice is in the TP calculation. The source declares input int InpTakeProfit = 20 and computes a local double tp = InpTakeProfit * _Point * 10, but that local is never passed to TryBuy or TrySell. The TP argument sent in the call is mid[0] — the BB middle band, recomputed every tick from CopyBuffer. The InpTakeProfit input is dead code. The reason is consistent with a mean-reversion philosophy: the EA is not trying to capture a fixed pip target, it is trying to ride the touch back to the statistical mean. On a ranging day mid[0] could be 50 pips away; on a trending day it could be 5 pips or 200 pips. The fixed 30-pip SL is the safety net; the TP is the equilibrium. There is no way to override this through the inputs — the InpTakeProfit value is purely cosmetic.

Position management is intentionally absent. There is no ManageTrailingStop() call, no break-even function, no partial close at 1R, no time stop, and no reverse-signal exit. A position opened by the BB touch closes only when the SL is hit, when price reaches the BB middle band, or when the broker force-closes (margin call, weekend gap, manual close, etc.). The OnTradeTransaction handler updates g_realizedToday, g_realizedWeek, g_tradesToday, and g_consecLosses from the deal stream — it does not move stops. The deliberate omission of in-trade management is what makes the 7-of-12 layer claim honest. EX09023 is for traders who want a one-shot BB fade with deterministic behaviour, not an adaptive exit scheme.

The 8-gate IsSafeToTrade_EX09023 walks in this strict order. (1) If InpCapAmount > 0 and equity >= InpCapAmount, return false — the soft dollar cap. (2) If InpCapFloor > 0 and equity < InpCapFloor, return false — the floor. (3) If g_initialBalance > 0 and equity < g_initialBalance * 0.95, return false — the 5% drawdown gate. (4) EffectiveCapital = min(CapAmount, equity) or full equity; if g_realizedToday <= -EffectiveCapital * 0.03, return false — the 3% daily loss limit. (5) Same check on g_realizedWeek at 2x (i.e., 6% weekly loss limit). (6) If consec losses >= 3 and current time < g_cooldownUntil (set 30 minutes after the third loss), return false — the cooldown gate. (7) If g_tradesToday >= 50, return false — the daily trade cap. (8) IsMarketOpen, InActiveSession, optional news filter. The session logic uses London 7-16 and New York 12-21 GMT, with an InpAvoidAsia toggle that blocks Asia-only bars (Asian hours are blocked only when London and NY are not also active, so the London-Asia overlap is still allowed). The news filter is time-based — 15 minutes around the London and NY session open hours, not the MQL5 economic calendar feed — and is OFF by default. DetectGMTOffset_EX09023 walks H1 bars for a 3+ hour gap and uses the gap to compute the server-to-GMT offset, falling back to InpServerGMTOffset if non-zero.

Order execution is a two-layer retry. The outer layer lives in OnTick: three attempts spaced 500ms apart, each with an OrderCalcMargin pre-check. If TryBuy or TrySell returns true, the tick returns. The inner layer lives inside TryBuy/TrySell: three attempts spaced 200ms apart, with a hard floor on retcode — only REQUOTE, TIMEOUT, PRICE_OFF, and PRICE_CHANGED trigger a retry; every other retcode hard-breaks. Lot is floor-rounded to the symbol's volume step, then clamped to [minLot, maxLot]. There is no deviation or slippage parameter, which means the EA uses the broker's default fill policy. On brokers with slow execution or wide requotes, the retry budget can be exhausted in milliseconds and a viable signal can be missed.

OnTester returns (STAT_PROFIT * STAT_PROFIT_FACTOR) / (1.0 + STAT_EQUITY_DDREL_PERCENT), with a 30-trade minimum and a guard against pf<=0 or dd<=0. This is a profit-per-unit-of-drawdown score; it rewards raw profit scaled by profit factor and penalised by drawdown. The 30-trade minimum is a hard cutoff — backtests with fewer than 30 closed trades return 0 and the optimizer will discard that parameter set. There is no multi-objective score and no walk-forward helper in the source.

In a backtest, expect 30-200 closed trades per symbol per year depending on the symbol's volatility and how often the BB outer band is touched. The XAUUSD M5 default is the natural fit: BB(20, 2.0) outer bands touch frequently on gold's intraday range, and the BB-middle dynamic TP gives the EA room to capture mean reversion on London and New York sessions. The 0.10 lot default and 30-pip SL are sized for a moderate account; with minDeposit $100 (per the metadata), most users will want to either reduce the lot or run a larger account. For traders who want a minimal BB fade with clean reporting, deterministic behaviour, and no surprise complexity, EX09023 is the family member to pick. For traders who want a pyramid, a regime switch, a multi-confirmation stack, or hedging-account-specific behaviour, look at EX09013, EX09015, EX09020, or EX09021 instead.