Pipsgrowth EX09024 MeanReversion

EX09024 RSIMeanReversion is a deliberate exercise in subtraction. Where most of the EX09 family composes two or three indicators into a multi-stage decision, EX09024 takes one oscillator — RSI with InpRSIPeriod=14 — and two threshold-cross rules. That is the entire signal stack. Buy when the previous bar's RSI was below InpOversold=25 and the current bar's RSI has crossed back to 25 or above. Sell when the previous bar's RSI was above InpOverbought=75 and the current bar has crossed back to 75 or below. The rest of the file is risk and safety scaffolding, not signal generation.

The cross requirement is what separates EX09024 from a plain "RSI < 25" oversold-buy. EX09024 only fires on the bar where RSI exits the extreme zone, not on the bar where RSI is still deep in it. A long signal needs rsi[1] < 25 AND rsi[0] >= 25 — both legs have to be true on the same pair of bars. A short signal mirrors that: rsi[1] > 75 AND rsi[0] <= 75. The 25/75 defaults are deliberately tighter than the conventional 30/70, which means fewer signals but a higher expectation that momentum has actually rolled over. The two RSI values are read straight from the buffer at OnTick: CopyBuffer(hRSI, 0, 1, 2, rsi) returns a two-element array of bar 1 and bar 0 in that order, and the code indexes rsi[1] and rsi[0] directly without copying to a struct.

Everything runs on a single indicator handle, hRSI, allocated in OnInit as iRSI(_Symbol, PERIOD_CURRENT, InpRSIPeriod, PRICE_CLOSE). OnInit refuses to start if the handle returns INVALID_HANDLE, if InpMagicNumber is zero, if InpLotSize is non-positive, if InpCapFloor >= InpCapAmount when a cap is configured, if InpMaxTradesPerDay is non-positive, or if InpMaxLot is non-positive — each with a Print() and a return of INIT_PARAMETERS_INCORRECT. The OnTick body is roughly fifteen lines of actual signal logic; the rest of the function is the TryBuy_EX09024 / TrySell_EX09024 retry wrappers.

Position management is the simplest the EX09 family offers. Once a position opens, EX09024 sets a hard stop at InpStopLoss = 40 × _Point × 10 = 400 points (on XAUUSD that is 40 pips / $4.00 at 0.10 lot) and a hard take-profit at InpTakeProfit = 40 × _Point × 10 = 400 points. The R:R is 1:1 by design. There is no trailing stop, no break-even ratchet, no partial close, no time-based exit, no opposite-signal close, no basket TP. The exit is whichever side hits first: SL or TP. EX09024 is a strategy that depends on the win-rate of the entry, not on money management after entry.

Lot sizing is fixed at InpLotSize = 0.10. TryBuy_EX09024 and TrySell_EX09024 both clamp the requested lot to the symbol's min/max/step — MathFloor(lots/step)*step, then floored to minLot and capped to maxLot — and both pre-check margin via OrderCalcMargin against ACCOUNT_MARGIN_FREE before the first send. If margin is insufficient, the function returns false immediately with a print, and the outer 3-attempt loop never runs. The retry loop itself is 3 attempts on REQUOTE, TIMEOUT, PRICE_OFF, or PRICE_CHANGED at 200ms Sleep intervals; any other retcode is a hard fail and the loop breaks. The order is sent with CTrade::Buy / CTrade::Sell, which uses the broker's default filling mode via the CTrade class (FOK, IOC, or RETURN are broker-dependent).

EX09024 enforces single-position discipline through two distinct layers. The first is HasTrade(), a scan of PositionsTotal() that returns true if any open position has POSITION_MAGIC matching InpMagicNumber=22209024 and POSITION_SYMBOL matching _Symbol. If HasTrade() is true, OnTick returns immediately and no new entries can fire while a position is open. The second is a one-trade-per-bar guard via the module-level lastBar datetime. On every tick, OnTick reads iTime(_Symbol, PERIOD_CURRENT, 0) and compares to lastBar; if the current bar timestamp equals the last processed bar, the tick is skipped. The bar gate means at most one entry attempt per new bar even if multiple valid cross signals somehow form within the same candle. There is no time-stop on the open position — if the candle does not exit the extreme by the time the bar closes, the signal is gone, but the open trade still has to wait for SL or TP.

The IsSafeToTrade_EX09024 gate is the full EX09 hardening stack, applied as nine sequential checks. (1) Capital cap reached — if InpCapAmount > 0 and equity >= InpCapAmount, no trading. (2) Capital floor — if InpCapFloor > 0 and equity < InpCapFloor, no trading. (3) Initial equity guard — if equity < 95% of g_initialBalance, no trading. (4) Daily loss limit — realized P&L <= -3% of effective capital. (5) Weekly loss limit — realized P&L <= -6% (3% × 2) of effective capital. (6) Consecutive-loss cooldown — if g_consecLosses >= 3, block for 30 minutes from g_cooldownUntil. (7) Daily trade cap — g_tradesToday < 50. (8) Market-open check — closed Friday after 21:00 GMT, all of Saturday, Sunday before 22:00 GMT. (9) Session filter — only London 7–16 GMT and New York 12–21 GMT count, and Asia 0–7 GMT is explicitly avoided when InpAvoidAsia is true (the default). An optional news filter, off by default (InpNewsFilter=false), blocks ±15 minutes around the London and NY session open hours if enabled.

The on-trade bookkeeping uses OnTradeTransaction to scan DEAL_ADD events. For every new deal, the code selects the deal, checks DEAL_MAGIC matches this EA's magic and DEAL_ENTRY = DEAL_ENTRY_OUT, and accumulates DEAL_PROFIT + DEAL_SWAP + DEAL_COMMISSION into g_realizedToday, g_realizedWeek, and g_tradesToday. If the deal P&L is negative, g_consecLosses increments and g_cooldownUntil is set to TimeCurrent() + 30 × 60 seconds. UpdateDailyCounters_EX09024 runs at the top of every tick and resets g_tradesToday, g_consecLosses, g_realizedToday when the calendar day rolls over (compared via StringToTime on a year.month.day string). Weekly is not auto-reset in the same loop — it is reset implicitly when the strategy tester or terminal restarts the EA, and the limit applies to the running week only.

GMT handling is via DetectGMTOffset_EX09024, which walks H1 bars looking for a >2-hour gap (the weekend), then computes the offset as (hour of first post-gap bar) - 22, normalized to [-12, 12]. If InpServerGMTOffset is non-zero, that hardcoded value overrides the auto-detect. ServerToGMT_EX09024 subtracts g_gmtOffset × 3600 from the server time, and all session checks operate on that converted time. This is what makes the London 7–16 / NY 12–21 logic correct across brokers with different server timezones (IC Markets GMT+0, GMT+2 ECN, GMT+3 Pepperstone, etc.).

The OnTester formula is the same shape as the rest of the EX09 family: net profit × profit factor / (1 + relative equity DD %), with a 30-trade minimum and a guard that returns 0 if profit factor is non-positive or relative drawdown is non-positive. For strategy tester genetic optimization, the inputs that actually move the result are InpRSIPeriod, InpOverbought, InpOversold, InpStopLoss, InpTakeProfit, and InpLotSize. The remaining 18 inputs are session, risk-cap, and hardening controls that don't change between optimization runs unless you re-test across brokers.

Honest read on the 12-layer header: the comment block claims REGIME, SIGNAL, ENTRY, CONFIRM, NO-TRADE, CAPITAL CAP, RISK, SIZING, MANAGE, EXIT, SCALING, OnTester — but in this file, only SIGNAL, ENTRY, NO-TRADE, CAPITAL CAP, RISK, SIZING, EXIT (via the static SL/TP at order send), and OnTester have actual code. There is no regime classifier (no ADX, no trend filter, no HTF check), no CONFIRM module (the cross IS the confirm), no MANAGE module (no BE, no trail, no partial), and no SCALING module (no pyramid, no martingale, no grid, no adds). This is a deliberate choice — EX09024 is the EX09 family member for traders who want the signal in plainest possible form, paired with the full safety/risk stack.

What to expect in backtest: XAUUSD on M5 with a 1:1 R:R depends on the bar-structure of gold. Gold's typical M5 range during London–NY overlap runs roughly 30–80 points; a 40-point stop is roughly half an average M5 range, so the SL will get hit more often than the TP on a single random distribution, and the strategy needs the RSI cross signal to be selective enough to push win-rate above 50% to clear breakeven after spread. On H1, the same 40-point SL/TP becomes much smaller relative to a typical 200–400 point range, which favors the strategy — the 1:1 R:R becomes a smaller ask against a wider distribution of outcomes. Brokers with sub-point XAUUSD spreads (ECN or RAW) materially help because the entry happens at a level where RSI is already bouncing, and a wide spread eats into the room. $100 micro-lot, 0.10 fixed, runs cleanly on the minimum deposit.