Pipsgrowth EX16029 Trend

EX16029 is a 2-bar inside-bar reversal EA built around the classical Japanese Harami candlestick pattern, filtered by a 5-period SMA trend context and confirmed by a 37-period Money Flow Index. The Harami is the small-bodied bar that forms inside the prior bar's body — a containment setup that signals indecision at the end of a directional move. EX16029 treats that containment as a reversal cue only when the prior move has just delivered a long-bodied candle and the broader tape context still agrees.

The Harami geometry is enforced by four body-relationship conditions plus one trend gate. For a Bullish Harami, the last completed bar must be bullish (Close(1) > Open(1)) while the candle two bars back must have been a long bearish body — long meaning its body size exceeds the 12-bar average body length tracked by AvgBody(1). The bullish bar's body must then sit inside the prior bearish body: its close below the prior open AND its open above the prior close. The fifth condition is the trend gate: the midpoint of the prior bar's full range — MidPoint(2) = (High(2)+Low(2))/2 — must sit below the 5-SMA close value, which proves the market was already trending down before the inside-bar formed. A Bullish Harami that forms mid-trend against a downtrend context is treated as a more meaningful reversal. The Bearish Harami mirror is symmetric: last bar bearish, prior bar a long bullish body, current bar's body inside the prior body's range, and MidPoint(2) above the 5-SMA to confirm uptrend context. The use of the range midpoint (High+Low)/2 rather than the body midpoint (Open+Close)/2 is deliberate — it pulls the wicks and shadows into the trend check, so a long lower wick on the prior candle (a hammer-like rejection footprint) softens the downtrend signal even if the body sits well above the SMA.

Confirmation is delegated to a 37-period MFI applied to tick volume (VOLUME_TICK). For a buy the EA requires MFI(1) < 40, for a sell MFI(1) > 60. A 37-period MFI on M5 bars looks back roughly 3 hours of price-volume action — long enough to integrate the cumulative tape rather than react to a single spike, short enough to stay current with the active session. The 40/60 bands sit just inside the classic 20/80 overbought/oversold thresholds, which keeps the filter relatively permissive: a Harami that forms while the MFI is merely tilting toward exhaustion (rather than pinned to a deep extreme) still passes through. Without confirmation the signal is dropped silently — ExtSignalOpen is reset to SIGNAL_NOT before any order is submitted.

Risk and exit are tightly bound to the entry. Every position is opened with a 200-point stop and a 200-point target for an exact 1:1 reward-to-risk ratio. Both orders carry a spread-override safety net: if the current spread is wider than the SL or TP distance in points, the EA clamps the protective level to the spread itself rather than placing a stop inside the spread. The lot size is fixed at 0.10 lots and there is no per-trade risk-percent or balance-proportional sizing. A 10-bar time stop is also wired in via BarsHold(), which counts completed bars since the position opened; once the bar count hits the configured InpDuration=10, the EA force-closes the position regardless of P&L.

The exit-side MFI logic adds an early-close layer on top of the 200-point target. Two close patterns are recognised. For long positions, an exit fires when MFI(1) < 70 AND MFI(2) > 70 (overbought drop through the 70 line) OR when MFI(1) < 30 AND MFI(2) > 30 (oversold recovery bounce — the MFI failed to hold the deep oversold level). The short mirror is the same logic flipped at the 30 and 70 thresholds. These cross signals are checked once per new bar via the next_bar_open static datetime guard in OnTick(), which means the EA never re-evaluates exits more than once per period — a deliberate throttling that prevents whipsaw closes from intrabar noise.

The MFI indicator is read through iMFI(_Symbol, _Period, 37, VOLUME_TICK) and the trend SMA through iMA(_Symbol, _Period, 5, 0, MODE_SMA, PRICE_CLOSE). Both handles are created in OnInit. A small bug in OnInit checks the MFI handle variable a second time when validating the MA handle — line 95 reads if(ExtIndicatorHandle==INVALID_HANDLE) where the code should test ExtTrendMAHandle. The MA init is therefore protected by accident only when the MFI handle is invalid; in normal operation the MA always succeeds so the bug is invisible. A second, more visible leak appears in OnDeinit: only the MFI handle is released, and the SMA handle is dropped on EA unload without an explicit IndicatorRelease(ExtTrendMAHandle). For traders who reload the EA frequently on the same chart, this leaks a handle each time until the terminal's indicator slot is exhausted.

Three retry helpers are defined at the bottom of the file — TryClose_EX16029, TryClosePartial_EX16029, and TryModify_EX16029 — each looping up to three times on REQUOTE/TIMEOUT/PRICE_OFF/PRICE_CHANGED codes with 100-200ms sleeps between attempts. Only the first is wired into the live close path; TryClosePartial and TryModify are dead code, kept as scaffolding for partial-close and stop-ratchet features that the EA never enables. There is no trailing stop, no break-even shift, no basket-level close, and no equity-curve gating — the EA is intentionally minimal: one indicator pair, one pattern family, one fixed lot, one RR.

The target deployment is XAUUSD M5 by default, with H1 as a viable alternative for traders who prefer slower candles. The strategy depends on a low and stable spread to keep the 200-point SL/TP meaningful — on gold, 200 points is $0.20/oz on a 2-decimal broker or $2.00/oz on a 3-decimal broker, which translates to roughly 0.7%–1.5% account risk on a $1,000 balance at 0.10 lot. ECN/RAW accounts with sub-pip typical spreads are the natural fit; standard accounts with 30-50 cent gold spreads will see frequent spread-override events and most trades will exit at the spread rather than the configured level.