Pipsgrowth EX09012 MeanReversion

Pipsgrowth EX09012 is a one-indicator reversal system. Where the rest of the MeanReversion family in this catalogue layers candlestick geometry, volume-trap detection, CCI confirmation, or composite scoring, EX09012 stays narrow: a single Relative Strength Index on the closed bar, and a binary crossover against the 30/70 oversold and overbought levels. The strategy buys when the most recent fully-closed RSI(14) crosses up from below 30 — i.e. a bar that closed oversold followed by a bar that closes back inside neutral — and sells on the mirror: a bar that closed above 70 followed by a bar that closes back below 70. The thresholds and the lookback period are user-configurable via InpRSI_Period, InpRSI_Oversold and InpRSI_Overbought, but the default 14/30/70 combination is the canonical RSI reversal setup.

The signal is detected on a new-bar gate inside CheckSignal(). The function reads two RSI values from the last closed bar — bufRSI[0] is the most recent closed value, bufRSI[1] is the one before that — and only fires once per bar via the static g_lastBarTime compare. That gate is what keeps the EA from re-entering the same signal on every tick of a flat bar. Once the direction is decided, OpenPosition() pulls the live ask or bid, computes SL and TP as fixed-point distances, runs an OrderCalcMargin pre-check against free margin, and then attempts the order through a 3-retry loop on TRADE_RETCODE_REQUOTE, _PRICE_CHANGED, _PRICE_OFF and _TIMEOUT, with 100ms sleeps and a RefreshRates() between attempts. The fill mode is resolved once in OnInit by SetProperFillingMode(), which probes the symbol's SYMBOL_FILLING_MODE and picks FOK, IOC or RETURN in that priority order.

Stops are fixed points, not ATR multiples. The default configuration is 500 points for the stop-loss and 1000 points for the take-profit, both computed as price ± points * mysymbol.Point() and rounded to the symbol's digits. At 2:1 reward-to-risk the math is simple — a winning trade captures twice the loss of a losing one — but the distance is constant regardless of how quiet or volatile the market is. A 200-bar quiet session in XAUUSD will get the same 500/1000 point treatment as a 200-bar breakout session, which is by design: this is a point-based EA, not a volatility-aware one. The 5-lot cap (InpMaxLotSize) is what keeps a mis-set InpRiskPercent from over-sizing on a thin account.

What EX09012 does have that simpler RSI bots do not is a two-stage trade-management stack running on every tick in ManageOpenPositions(). Stage one is the break-even: when the position is InpBEStart=250 points in profit, the SL is moved to entry + InpBEOffset=30 points for a long, mirrored for a short. Stage two is the trailing stop: once the position is InpTrailStart=350 points in profit, the SL ratchets to price - InpTrailDistance=150 points behind the current bid (or ask for shorts), and only ever moves forward. The two stages are designed to stack — a 500-point move first locks in break-even-plus-30, then the trailing stop takes over and follows price at a 150-point leash. There is no partial close at 1R like the other MeanReversion EAs use; the position either runs to the 1000-point TP, gets stopped at break-even-or-trail, or hits the 500-point SL on the way out. Both stages call TryModify_EX09012, which re-attempts the SL change 3 times on REQUOTE and TIMEOUT only with 100ms sleeps between.

The risk layer is more thorough than the entry logic. CheckDrawdownLimits() runs on every tick and tracks two numbers: a daily start balance and a running peak equity. If realized daily loss reaches 3% of the start-of-day balance, or if current equity falls 10% from the running peak, g_ddLimitHit is set to true, the EA optionally closes all positions via CloseAllPositions() and no new entries are taken until CheckNewDay() resets the flag at the next midnight server time. The peak-equity logic matters because it catches flash drawdowns that the daily-loss check would miss — a 9% mid-day dip counts as peak - 9%, not as dailyStart - 9%, and the 10% threshold means a single bad session closes the EA before it can dig deeper. InpCloseOnDDLimit=true (the default) means the EA force-closes open positions when the limit fires, not just refuses new entries.

Beneath the basic time filter (InpUseTimeFilter=false by default, 8–20 server hours, Friday close at 20) sits a second-layer hardening module that most MeanReversion EAs in this catalogue do not expose as input. DetectGMTOffset_EX09012() walks back through hourly bars looking for a daylight-saving gap to derive the server's GMT offset; if InpServerGMTOffset is set non-zero the user override is used instead and detection is skipped. InActiveSession_EX09012() then converts server time to GMT and checks whether the current minute is inside the London (InpLondonStartHour=7 to InpLondonEndHour=16) or New York (InpNYStartHour=12 to InpNYEndHour=21) windows, with an InpAvoidAsia block on the 0–7 GMT band. The session gate is what actually decides entry eligibility — the older 8–20 server-hour filter only matters if you set InpUseTimeFilter=true, but the GMT-aware session is always on. IsNewsTime_EX09012() is a generic time-based proxy that flags the ±InpNewsFilterMinutes=15 window around the London and NY session opens. None of this is a real news feed — it is a coarse time fence, off by default, useful only as a sanity filter against the first 15 minutes of a major release.

A capital-allocation cap and floor round out the safety stack. Set InpCapAmount=0 to disable, or set it to your dollar risk budget and IsSafeToTrade_EX09012() will treat MathMin(cap, equity) as effective capital for both lot sizing and the daily-loss check. InpCapFloor=50 blocks entries below $50 equity. InpMinEquityPercent=95 blocks entries when equity has fallen more than 5% from the g_initialBalance captured in OnInit. InpMaxTradesPerDay=50 caps the daily trade count, and InpMaxConsecLosses=3 plus InpCooldownMinutes=30 enforces a 30-minute cool-off after a third consecutive loss — the consecutive-loss counter is incremented inside OnTradeTransaction on every losing DEAL_ENTRY_OUT. None of these are entry filters on signal quality; they are the EA's only defense against a long losing streak.

Pyramiding is wired in but disabled by default. Flip InpEnablePyramid=true and IsGridSafe() will allow a second or third position in the same direction, capped at InpMaxPositions=3, but only if the existing same-direction position is already at least InpMinProfitPoints=200 points in profit. That is the gate: a layered position can only be added when the first layer is green, which prevents the classic grid-martingale pattern of averaging into losers. The function counts same-direction positions in CountPositions() and refuses the pyramid entry if any existing position is in the red. There is no lot multiplier on the pyramid leg, no hedge-against-opposite, and InpEnablePyramid=false keeps the EA flat in one position at a time.

What the EA does not have is worth listing. There is no regime classifier — unlike the EX09004 volume-trap and EX09006 candle-pattern EAs in the same family, EX09012 will fire an RSI cross on a strong trend and let the fixed SL handle the loss. There is no HTF (higher-timeframe) confirmation, no candle-pattern gate, no volume gate, and no volatility gate on the signal bar. There is no partial close and no trailing TP — TP is hit once at the 1000-point level, the SL/TP ratio is the entire edge. The OnTester custom criterion is (net * profitFactor) / (1 + relativeEquityDDPercent) with a 30-trade minimum before the score is returned. That formula penalizes high-drawdown optimizations, which means an MT5 strategy-test pass that has higher net profit but deeper DD will score worse than a smoother alternative with lower net. The optimizer will tend to favor parameter combinations that do not blow up, not the ones that make the most.

For backtesting expectations: on XAUUSD M5 with the default 500/1000 points, expect somewhere between 200 and 600 trades per year depending on the test window's RSI-cross frequency, a hit rate typically in the 40–55% range (RSI crosses are not a high-probability signal alone), and a profit factor in the 1.0 to 1.4 zone. The 2:1 fixed R:R is the only edge the strategy has against randomness; the break-even and trailing stack exist to keep that edge from leaking back. Trades that hit TP will land exactly at 1000 points. Trades that hit the 500-point SL will lose the full distance. Trades that get trailed will close somewhere in between, and the break-even floor means the worst-case after the position is +250 points in profit is a small win, not a loss. The DD protection and the 3% daily-loss cap are what keep an unlucky streak from compounding into a margin call — on a $100 account a 10% peak-to-trough equity drop is $10, which is two losing trades at minimum lot, and the EA will halt before it can chain a third.