Pipsgrowth EX09001 MeanReversion

Pipsgrowth EX09001 MeanReversion is a single-instrument RSI mean-reversion EA built around a deliberately conservative posture. The strategy is meant for XAUUSD on M5 or H1, and it ships from the package with InpDryRun = true — meaning it will print what it would do without sending live orders. To go live, the user explicitly sets InpDryRun to false. This is a strong design statement: the EA assumes the operator will validate the strategy on a demo or in the MT5 Strategy Tester first, and treats live trading as an opt-in.

The core idea is straightforward. On every tick, the EA computes an RSI(21) and looks for the bar where RSI has just crossed back inside the overbought or oversold band, defined by InpRSILevel (default 70, so overbought ≥ 70 and oversold ≤ 30). The crossover is checked on the closed bar (shift 1), not the forming one, so a signal cannot fire on noise. RsiSignal() returns +1 for a buy recovery (RSI2 ≤ 30, RSI1 > 30) or -1 for a sell rejection (RSI2 ≥ 70, RSI1 < 70), and stamps a confidence score based on how deep RSI was before the recovery. A 30-confidence floor is added so even shallow re-entries register; deeper dips score higher.

That raw signal is then filtered through a regime classifier. DetectRegime() reads ATR(14), ADX(14), and a 100-bar lookback of Bollinger Band width. From these it produces one of seven states: Choppy, StrongTrend, WeakTrend, Range, Compress, Expand, or Breakout. StrongTrend is identified when ADX ≥ 30 — and if it occurs alongside a compressed ATR percentile, the regime is labeled Breakout rather than StrongTrend. Compress fires when the ATR percentile is below 30 OR the current Bollinger width is below 60% of the rolling average. Expand fires when ATR percentile is ≥ 70. WeakTrend covers 20 ≤ ADX < 30. Range is the default low-volatility residual. The default of InpRequireRangeRegime = true means entries are restricted to Range, Compress, and WeakTrend — strong-trend and breakout regimes are deliberately excluded, on the assumption that a mean-reversion trade is statistically poor when the market is in a one-way move.

ConfirmEntry() then adds three more filters. The first is the HTF-EMA check (InpConfirmHTF = true by default). The EA evaluates a 50-period EMA on H1, regardless of the chart's own timeframe. Counterintuitively, the buy signal requires the H1 close to be below the H1 EMA, and the sell signal requires H1 to be above — i.e. the entry is a pullback into the trend, not a continuation. This is the classic fade-the-retracement setup, and it presupposes that the H1 trend is the dominant force that the M5 RSI has just over-extended against. If the HTF is already on the same side as the signal, the trade is skipped. The second filter is volatility sanity: if the current ATR percentile is at an extreme, the trade is blocked. The third is a hard spread ceiling: spreads above MAX_SLIPPAGE_POINTS (30 points) are rejected outright.

Risk is fixed low: 0.5% of effective capital per trade. Effective capital is the lesser of account equity and an optional hard cap (InpCapitalCapAmount); if neither is set, the full equity is used. CalcLotByRisk() converts the SL distance in price to ticks, looks up the broker's tick_value and tick_size, and rounds the lot size down to the symbol's lot step. Position sizing is therefore mechanical and broker-aware: a 2.0×ATR SL on XAUUSD at 30-pip ATR produces a different lot on a 5-digit gold broker than on a 2-digit one.

Trade management runs every tick. The full SL is 2.0×ATR, the full TP is 1.5×ATR. That produces a 0.75 R:R target — sub-1:1 by design, betting on high win rate rather than large winners. A partial-TP mechanism is layered on top: when price reaches 1.0×ATR in profit, half the position is closed at the partial-TP level. The remainder continues to the full TP. Break-even is set when price reaches +1.0×ATR in favor, moving SL to entry. An ATR-based trailing stop (TRAIL_ATR_MULT = 2.5) ratchets the SL forward, but only if the candidate is ahead of the previous SL and ahead of entry — it never loosens. If the trade is in profit but hasn't hit the partial yet, a time stop fires after MAX_BARS_IN_TRADE (240 bars — 20 hours on M5, 10 days on H1). There is also a per-day realized-loss cap of 3%, a weekly cap of 6%, and a 5-bar cooldown after three consecutive losses. The session filter is 7:00 to 20:00 broker server time, which on a typical ECN broker corresponds to the open of the European session through the close of the US session.

What this EA does NOT do: it does not scale in (InpMaxConcurrent = 1, InpMaxSymbolExp = 1, no martingale, no grid). It does not run news filters or detect broker-side server time vs. UTC drift — the 7-to-20 window is interpreted on whatever TimeCurrent() reports. The OnTester() function returns (net × profit_factor) / (1 + max-DD%), with a 30-trade minimum — this is a quiet but important detail for anyone running the Strategy Tester, because the EA will not optimize well on backtests with fewer than 30 trades per pass.

The 20-input parameter set is split into six groups (Identity, Risk & Sizing, Capital Allocation Cap, Signal & Regime, Exit / Manage, Operational). The most consequential toggles are InpDryRun (default true), InpRequireRangeRegime (default true), InpConfirmHTF (default true), and InpOppositeSignalExit (default false). Flipping the last one to true adds an early exit when the opposite RSI signal fires mid-trade — useful in strong reversals, but the trade-off is more whipsaw closes in trend days.

For backtest expectations: on XAUUSD M5, this strategy trades sparsely — one entry per bar, but a high bar rejection rate. Expect a smooth equity curve with small average winners and small-to-moderate losers, more comparable to a market-making profile than a directional trend system. On H1, the same logic fires far less often, with proportionally larger per-trade swings.