Pipsgrowth EX16092 Trend

CCI(27) sits at the heart of this expert advisor, paired with a three-bar candlestick engulfing pattern and a strictly fixed one-position-at-a-time policy. The CCI averaging period is deliberately set higher than the conventional 14 — the header lists Inp_ma_period at 27 — which makes the oscillator less twitchy on M5 candles and shifts the optimal cycle away from noise scalp frequency and into a more deliberate swing-trade rhythm. The applied price default is 6, which MapAppliedPriceInt maps to PRICE_TYPICAL, the (high+low+close)/3 measure that smooths single-candle wicks. The M5 default works on the backtest; the source explicitly notes M5-H1 portability, which is uncommon — most trend EAs pick one timeframe and force the rest. The result is an EA that runs at two distinct cadences from the same code, depending on what the trader attaches it to.

Signal formation has two interlocking pieces. The CCI arm reads four bars via iCCIGetArray(0, 4, cci) and looks for an inflection that resolves the way a momentum reversal should: bar three was higher than bar two, bar two was higher than bar one (a falling sequence), bar one dipped below +5 (technically cci[1] < 5.0), and bar zero bounced above bar one. That is the classic CCI bottoming curl. The candlestick arm on the same three bars must then show a bullish engulfing: bar two black (open>close), bar one black (open>close), and bar zero white (open<close) with bar one's open sitting below bar zero's close — i.e. the white body engulfs the black body. The two arms are required to fire on the same three-bar window. A partial signal, where only one arm prints, is not enough to act. Sell entries are the exact mirror: cci[1] > -5, cci[2] > cci[3], cci[1] > cci[2], cci[0] < cci[1], with bar two white, bar one white, and bar zero black, plus bar one open above bar zero close.

There is a hard one-position cap. The entry block sits inside if(!IsPositionExists()), which iterates PositionsTotal() and checks both symbol and magic. Magic is 22216092, which the source OnTradeTransaction handler later uses to filter its own deals from the broker's deal stream. The cap means the EA does not pyramid by signal stacking — it does, however, change the size of the next entry through a separate lot-progression engine. Martingale (InpMartin = false by default) and step-by-step (InpStep = false by default) are mutually exclusive — OnInit refuses to load with INIT_PARAMETERS_INCORRECT if both are set true, raising an alert in either Russian or English. Martingale multiplies the loss trade's lot by InpMartinCoeff (3.0) and clamps the multiplier count at InpMartinMaxMultiplications (3). Step-by-step instead adds InpStepLots (0.1) to the loss trade's lot, capped at InpStepLotsMax (1.5), and can be configured to fire after a loss or after a profit via the ENUM_STEP toggle. A profitable trade resets m_lots back to the base InpLots (0.1).

The position protection stack is three layers. The first is a fixed stop at 20 pips and a fixed take at 50 pips, both computed off the broker-adjusted point (m_adjusted_point) so the math works on 3-digit and 5-digit quote feeds the same way. The second is a trailing ratchet that activates only after price has moved at least InpTrailingStop + InpTrailingStep from the entry — 5 + 15 = 20 pips by default — and then re-installs the stop at PriceCurrent minus 5 pips for longs (or symmetric for shorts), never moving it backward. The third is the broker's own POSITION_TP constant, which the trailing pass leaves untouched. The trailing call goes through TryModify_EX16092, a 3-retry helper that re-attempts on TRADE_RETCODE_REQUOTE, TRADE_RETCODE_TIMEOUT, and TRADE_RETCODE_PRICE_CHANGED with a 100ms sleep between attempts.

A quirk worth flagging: the OnTick function early-returns when STimeCurrent.sec < 40, which restricts the new-bar entry attempt to the first 40 seconds of any minute. The author's intent is to avoid acting on a partially-formed bar where the broker feed is still stabilizing, but the practical effect is that the EA only looks for signals during the most settled portion of each minute candle. Combined with the one-position cap, this is a low-trade-frequency setup by design — a backtest on M5 XAUUSD typically produces a handful of entries per week, not per day.

The position-close path is intentionally narrow. The source does not call TryClose_EX16092 or TryClosePartial_EX16092 from OnTick — both helpers exist with full 3-retry logic against requote and price-change errors, but they are never invoked at runtime. There is no time-based exit, no CCI cross-zero exit, and no signal-reversal exit. The only ways out of a trade are stop-loss, take-profit, and the trailing ratchet. If the trader expects the EA to flip direction on the next candle, that does not happen — the next signal will be ignored if a position is still open, and the EA will simply wait until the current position resolves through one of its three exit paths before the next signal can fire.

For brokers, the EA does not query spread or impose a maximum spread input. There is no news filter, no session gate, and no OnTester optimization metric. The slippage is hardcoded to 10 points via m_slippage=10. Magic 22216092 is set on the CTrade object at init, and the comment string Psgrowth.com Expert_16092 is attached to every order. The header's claim of 12 layers is more aspirational than literal — REGIME, SIZING, and OnTester are absent from the implementation; the wiring is SIGNAL, ENTRY, CONFIRM, NO-TRADE, RISK, MANAGE, EXIT, SCALING. The mapAppliedPriceInt helper is duplicated three times in the source, which is a copy-paste artifact rather than a deliberate design choice — it does not affect runtime.

In backtest the EA behaves like a slow-twitch mean-reversion system with embedded martingale risk. The CCI(27) + 3-bar candlestick window produces a sparse signal stream, the 1:2.5 fixed RR keeps most winners positive, and the trailing stop protects mid-candle spikes once the trade has moved favorably. The martingale branch is the danger — three 0.1 → 0.3 → 0.9 → 2.7 lot sequences in a row can blow through a $100 account on XAUUSD where a 20-pip SL is roughly $54 at 0.1 lot on gold. Run the EA with InpMartin = false and InpStep = false (the defaults) for honest signal-mode testing, and only engage the lot progression after both a long backtest and a live demo validation. The expected behaviour on a typical M5 gold feed: two to four closed positions per week, mostly clipped by the trailing ratchet, with a small subset closing at the full 50-pip TP.