Pipsgrowth EX09003 MeanReversion

Pipsgrowth EX09003 CycleSwing is a counter-cyclical XAUUSD EA that fades Commodity Channel Index extremes after a confirming Stochastic turn. The thesis is mechanical: when CCI has pushed beyond its oversold band (-100) or overbought band (+100) and the Stochastic main line crosses its signal back inside the middle band (below 50 for longs, above 50 for shorts), the cycle is assumed to be reverting, and the EA enters in the direction of the snap-back. With magic 22209003, version 2.00, and the suggested pair being XAUUSD on M5, the EA is purpose-tuned for a fast-fade model on gold where the bulk of profitable cycle reversals happen in 1-3 hour bursts.

The signal layer is the CycleSwingSignal() function, which reads two closed-bar values of CCI(14) and Stochastic(14,3,3) and checks for a band-cross plus a main/signal cross in the same direction. The depth of the prior excursion feeds a confidence value between 10 and 100 — a deeper CCI excursion under -100 before the cross yields a higher reading, on the logic that the more stretched the cycle, the stronger the snap-back. Buy and sell are exact mirrors: a long requires cciPrv < -100 && cciNow > -100 together with stochMain crosses above stochSignal while stochMain < 50; a short requires the opposite around +100. Reads on the live bar (shift 0) are not used — the EA only acts on fully closed bars, which avoids the most common source of false CCI re-entries.

Before any signal can fire, it has to clear a 7-state regime classifier. The ClassifyRegime() function uses ADX(14), ATR(14), and Bollinger Bands(20, 2.0) to label the current market as one of StrongTrend, WeakTrend, Range, Breakout, Compress, Expand, or Choppy. The classification depends on three things: ADX relative to 18 (chop) and 25 (trend), the ATR percentile over a 50-bar window, and the current Bollinger width against its recent average. For mean-reversion, only Range, Compress, and WeakTrend are tradeable. StrongTrend, Breakout, Expand, and Choppy all block entries by default — the explicit comment in ConfirmEntry() is "avoid fighting strong trend." This filter is the single largest determinant of when EX09003 trades at all; in a trending week, expect very few signals.

There is also a higher-timeframe confirmation. The EA pulls a 50-period EMA on H1 and refuses to enter if price is more than 3 × ATR(14) away from that EMA. The point is to allow counter-trend scalps on mean-reversion pairs only when the broader structure is not actively trending — price hugging the H1 EMA is a permissive condition, while price 3 ATRs above the EMA in a 1H uptrend is a no-trade. If you disable the regime filter (the InpUseRegimeFilter input), only the Choppy regime still blocks. The session window is opt-in via InpUseSessionWindow and, when enabled, restricts entries to broker-server hours 06:00-20:00, which roughly covers the Asia→London handover into the New York pre-open. It is off by default.

The no-trade gate adds a half-dozen mechanical checks. A kill switch (InpKillSwitch) halts entries instantly. Spread must be under 400 points (the cap is XAUUSD-tuned — at 2-decimal gold, 400 points is $0.40, so a wide retail spread will gate entries). A realized-daily-loss check sums DEAL_PROFIT+SWAP+COMMISSION for closed deals with this magic today and refuses new entries once the cumulative realized loss hits 3% of effective capital. After a losing close, a 5-bar cooldown (≈25 minutes on M5) blocks re-entry. A capital floor of $50 stops the EA from trading on depleted accounts. Weekend hours are blocked. A high-importance news check uses CalendarValueHistory against the two currencies of the chart symbol (XAU and USD), and refuses entries inside a ±30-minute window around red-flag events. Finally, a SYMBOL_TRADE_MODE_FULL check makes sure the broker has not disabled trading on the symbol.

Position sizing is risk-percent driven. With InpRiskPercent = 0.5 as the default and the 1.5 × ATR(14) stop distance, the EA computes the risk per lot from tick_value and tick_size, divides 0.5% of effective capital by that risk, and falls back to 0.01 lot if either side returns zero. A 200-point hard floor (MIN_SL_POINTS) is applied to the SL distance before sizing so the lot cannot blow up in a low-ATR environment. Effective capital itself is governed by an opt-in cap: setting InpCapCapAmount > 0 means the EA pretends the account only has that many dollars for risk purposes, which is useful for staged scaling.

Once in, management is handled by ManageOpenPositions(), which runs every tick. Six distinct events are handled: (a) a one-shot break-even at +1R, moving the stop to open + 1 point once price has travelled one ATR-multiple in favor; (b) a one-shot partial close at +1R, offloading 50% of the position (only if the remaining volume is at least the broker's minimum lot); (c) an ATR trail at 2.0 × ATR(14) behind price, only ratcheting forward; (d) an opposite-signal exit that closes the trade if a fresh signal in the other direction fires; (e) a time-based exit at MAX_BARS_IN_TRADE = 288 bars (24 hours on M5); and (f) a regime-choppy exit that closes the position immediately if the classifier transitions to Choppy. All three retry helpers (TryClose_EX09003, TryClosePartial_EX09003, TryModify_EX09003) re-attempt on REQUOTE/TIMEOUT/PRICE_OFF/PRICE_CHANGED with a 200ms sleep (100ms for modify).

The default state out of the box is dry-run: InpDryRun = true means the EA logs every signal, every regime label, every gating decision, and every simulated fill, but does not send orders. The intended workflow is to attach the EA to a chart, watch the verbose log for several days across different sessions and regimes, confirm that the trade frequency, regime gating, and SL/TP distances match expectations, and then flip InpDryRun to false. The EA also ships with a custom OnTester() criterion: net * profit_factor / (1 + relative_drawdown), which means the Strategy Tester will rank parameter sets that combine net profit and profit factor while penalising drawdown, with a hard floor of 30 trades and a positive PF before the score is non-zero.

What this EA is not: it is not a trend-follower, not a grid, not a martingale, and not a pyramider. MAX_PYRAMID_LEVELS = 1 and there are no lot multipliers. It is a tight, single-position, regime-gated cycle-fader designed for XAUUSD on M5, and its edge comes from the regime filter and the HTF EMA distance gate refusing trades in the wrong structural context. The minimum recommended balance is $100, with the broader expectation that anything below $250 will see the 3% daily-loss cap hit on a bad day before the EA's edge can play out across many trades. Pair it with an ECN or raw-spread account so the 400-point spread cap is rarely the gating factor.