Pipsgrowth EX03004 Grid

Pipsgrowth EX03004 is a one-direction, trend-pyramiding grid for USDJPY. It only buys. There is no short path anywhere in the code — no short entry function, no sell call, no close-on-short logic. The whole lifecycle is built around catching a bullish swing on JPY pairs, adding to it as price moves in favor, and dumping the entire stack the moment momentum cracks. That single-direction bias is not a default; it is the architecture, and it shapes every risk decision in the file.

The trend gate is a three-EMA stack at 8, 13, and 21 periods on the close. The IsBullishTrend() helper returns true only when EMA(8) sits above EMA(13) and EMA(13) sits above EMA(21) on the current bar — a strict ordering, not just a slope. The bearish-exit helper, IsReversalSignal(), watches for a single event: EMA(8) crossing below EMA(13) on the current bar while having been at or above EMA(13) on the previous bar. That one cross triggers CloseAllBuyPositions() and the EA returns for the tick. The 21-EMA does not gate entries directly, but its position under the stack filters out shallow bounces that would otherwise trigger the 8/13 cross too early.

The pyramid mechanic lives in ManagePyramidStrategy(), which runs on every tick when the trend is bullish and the safety stack clears. On a clean chart (no buy positions), the EA fires the initial entry: 0.10 lots by default (InpInitialLots), with a 30-pip initial stop below the ask. Once at least one buy is open, the function switches to pyramid mode. It walks the open positions, finds the highest open price, and only opens a new buy when the current ask is at least InpPyramidPips above that level (default 20 pips). A second guard: the most recent position must already be in profit. If the last pyramid is still red, the function returns and waits for the next qualifying tick. The pyramid stop is tighter than the initial one — ask minus 0.8 × pyramidDistance — so each new leg is protected closer to entry as the stack grows.

The break-even-secure step is what makes the pyramid workable. For every open buy, the function checks whether bid has cleared openPrice + 10 pips AND the current stop is still below the open. If both are true, it moves the stop to openPrice + InpSecurePips (2 pips above entry). That locks the position into a small guaranteed profit on the next adverse move. The 2-pip buffer is small enough to keep the trade from getting stopped out by a typical pullback into the prior candle's range, but large enough to cover spread + slippage in normal USDJPY conditions. Securing is one-shot — once a position is at BE+2, the EA does not ratchet the stop further; the pyramid's own new entries provide the next layer of protection.

Exit is intentionally minimal. There is no TP, no time-stop, no per-position partial close. The only closing path is the asymmetric EMA cross described above. The reasoning is built into the file name: aggressive pyramid. The EA is designed to catch a single directional move from start to finish, then reload on the next confirmed bullish stack. Holding through pullbacks is the whole point. The flip side is that drawdowns during a long pyramid can be substantial if the trend reverses sharply — the EA will sit in red through the entire unwind of the EMA stack, only closing once the 8 crosses below the 13.

The safety stack is layered. IsSafeToTrade_EX03004() blocks new entries when: outside the configured session (London 7-16 GMT ∪ NY 12-21 GMT when InpUseGMTSessions is true), inside a cooldown window (set to TimeCurrent() + 30 minutes after the third consecutive loss, when InpCooldownMinutes > 0), below the equity floor (InpEquityFloor), past the daily or weekly realized loss limit, above the capital cap, or having hit the daily trade ceiling (InpMaxTradesPerDay = 50). InpMaxConsecLosses is exposed but defaults to 0 (disabled) — the user has to set it explicitly for the consecutive-loss gate to fire. The session itself is computed through DetectGMTOffset_EX03004(), which does a brute-force search across offsets from -12 to +12, scoring each candidate by whether the resulting GMT hour lands on a weekday between 07:00 and 21:00, and the winner becomes g_gmtOffset.

Order execution is conservative. TryBuy_EX03004() runs a pre-trade margin check via OrderCalcMargin — if the required margin exceeds free margin, the buy is refused before any ticket is sent. The actual buy uses the trade object's SetTypeFilling(ORDER_FILLING_IOC) and SetDeviationInPoints(InpSlippage = 30), and retries up to three times at 200ms intervals if the broker returns REQUOTE, TIMEOUT, PRICE_OFF, or PRICE_CHANGED. TryModify_EX03004() and TryClose_EX03004() use the same retry pattern. There is no spread filter and no news filter in the source — a deliberate omission for a JPY-pair pyramid that wants to be in the market whenever the EMA stack permits.

The OnTester() fitness function is (profit / maxDD) × PF, but only if maxDD > 0 and at least 10 trades closed. PF is forced to 0 if it is at or below 1, so the optimizer is heavily biased toward parameter sets that produce a real edge, not raw profit on a lucky handful of trades. Because the EA has no fixed TP, the optimizer's job is to find EMA period combinations, pyramid distance, and secure pips that ride the longest trends and bail cleanly on the reverses.

Practical expectations. On USDJPY M5, the EA will typically fire 1-3 buys per trend, secure them to BE+2, and exit on the EMA cross. Pyramiding to the InpMaxPositions=10 cap is rare and requires a sustained 200+ pip move in JPY pip terms. Drawdown scales with pyramid depth — going from 1 to 5 open positions on a single reversal can leave a 100+ pip floating loss before the cross triggers. The minDeposit of $100 reflects the initial 0.10 lot sizing, but a serious deployment on JPY pairs with this pyramid depth should start with at least $500-$1,000 of headroom. The pair restriction is real: OnInit() prints a warning if the symbol doesn't contain 'JPY', and the pip value calculation in the source assumes 3- or 5-digit pricing. Plugging this EA into EURUSD without adjusting the pip math will produce 10x-off stops.