Pipsgrowth EX18094 TrendFollow

Pipsgrowth EX18094 is a one-indicator RSI scalper built around a simple, almost mechanical recovery-then-fade thesis. The whole EA watches a single iRSI(14, applied price = close) on the configured timeframe — H1 by default because the TimeFrame input defaults to 9, which falls through the MapTimeframeInt switch to PERIOD_H1 — and trades a single position at a time. There is no regime classifier, no HTF confirmation, no session filter, no spread gate, no news filter, and no pyramid. Every entry and every exit is decided from the same 14-period RSI buffer, and the EA's behaviour from one tick to the next is fully captured in 9 functions totaling roughly 400 lines of MQL5.

The entry rule is a two-bar cross of the oversold or overbought line. The defaults shipped in the source are RSI_Oversold = 55 and RSI_Overbought = 82, which is a noticeably wider pair of thresholds than the 46/49 pair seen on most RSI-recovery EAs. The buy arm fires when rsi_two_bars_ago <= 55 AND rsi_prev > 55 — i.e. RSI spent a closed bar inside the 55-or-below zone and then closed the next bar back above 55. The sell arm is symmetric: rsi_two_bars_ago >= 82 AND rsi_prev < 82. The two-bars-ago check is the only debounce in the entry path; it kills the case where RSI whipsaws across 55 on two consecutive ticks but never actually closed there on a previous bar. The CheckEntrySignals() function calls OpenBuyPosition() or OpenSellPosition() directly, which call trade.Buy(LotSize, _Symbol, ask, 0, 0, ...) and trade.Sell(LotSize, _Symbol, bid, 0, 0, ...) — every bracket is zero, so the broker records the position with no attached stop loss and no take profit.

LotSize is a fixed input and the source ships it at 50, which is roughly two orders of magnitude above what a $100 minimum deposit would normally support on a 100 oz XAUUSD contract. This is a parameter that the operator is expected to set per account size, not a recommended default for the as-shipped account size. There is no risk-percent sizing, no balance-based scaler, and no margin check beyond the broker's own. CTrade is configured with magic 22218094, slippage 3 points, and ORDER_FILLING_FOK; FOK means the order either fills at the full requested size or does not fill at all, which is worth knowing if you change LotSize to something your broker cannot fill in one shot.

Exits are managed inside CheckExistingPosition() and they are the more interesting part of the EA, because there are two completely separate exit paths and they have a precedence relationship baked in. Path 1 is the target exit. For a long, the trade closes when rsi_current >= RSI_Target_Buy; for a short, when rsi_current <= RSI_Target_Sell. The default target values are 39 for the buy and 35 for the sell. Both targets are BELOW the 50 line and well below the entry thresholds of 55 and 82 — meaning the EA does not exit on momentum extension, it exits on reversion. You buy a recovery cross out of 55 and close the position when RSI gives back down to 39, which is exactly the kind of "buy the dip, take the bounce, don't be greedy" logic that produces tight equity curves on ranging instruments. Path 2 is the bars-against timeout, governed by BarsToWait (default 2). If RSI moves back against the open position — long when rsi_current < RSI_Oversold=55, short when rsi_current > RSI_Overbought=82 — the EA starts counting consecutive bars where the indicator is on the wrong side. The moment RSI returns to the friendly side the counter resets to 0; if the count reaches BarsToWait (2 closed bars in a row) the position is force-closed by ClosePosition(). This is effectively a 2-bar invalidation: if the recovery thesis doesn't prove itself within two candles, the trade is dead.

There is no price stop. The 0/0 brackets passed to trade.Buy() and trade.Sell() mean the broker's stop-out margin rule is the only price-based protection on every open position. A 50-lot position with no attached SL on a $100 account is a margin-call candidate the first time the broker's maintenance threshold is breached, so the practical use of this EA almost always requires the operator to set LotSize to a value the account can actually carry, or to layer an external stop in the form of a manual intervention or a second EA that watches the magic number. The EA itself has no opinion about position risk beyond "close after 2 bars of RSI going the wrong way" and "close when RSI reverts past the target line."

Position tracking is done with two booleans and an integer ticket. On every new bar — gated by the last_bar_time check at the top of OnTick() — UpdateRSI() copies the latest three RSI values into a buffer, CheckExistingPosition() runs first, and CheckEntrySignals() only runs if position_open is false. There is no PositionSelect(_Symbol) sweep, no magic-number scan of the entire open-positions list, and no recovery path for a position that was closed externally — if the broker closes the trade for any reason, the EA's internal state (position_ticket, position_open, bars_against_count) is reset on the next tick when PositionSelectByTicket() returns false. Retry logic for the close itself is implemented in TryClose_EX18094: 3 attempts with a 200 ms Sleep() between retries, retrying only on TRADE_RETCODE_REQUOTE, _TIMEOUT, _PRICE_OFF, or _PRICE_CHANGED, and breaking on any other return code. Two unused-but-compiled helpers, TryClosePartial_EX18094 and TryModify_EX18094, exist for partial closes and SL/TP modifications; both are dead code in this build because no SL or TP is ever set, and no partial close is ever requested.

OnTick() also short-circuits with two early returns: a Bars(_Symbol, g_TimeFrame) < RSI_Period + 2 guard that prevents the EA from running before enough history exists for the indicator to produce stable values, and a same-bar return inside the new-bar gate that prevents the EA from re-evaluating the same closed bar twice in succession. There is no Print() logging, no info panel, no OnTester() override, no OnTrade() callback, and no chart-object commentary. The EA is intentionally quiet — operators who want trade-level diagnostics need to switch on the strategy tester journal or wrap the EA in an external logging harness.

The MT5 package itself is the standard Pipsgrowth drop: one .mq5 file, the standard Trade.mqh include, and 12 inputs split into a one-row Identity group and a Strategy Settings group containing the timeframe, RSI period, applied-price selector, four RSI thresholds, the bars-to-wait integer, the LotSize double, the magic, and the slippage. A known source-level artifact is that MapTimeframeInt() and MapAppliedPriceInt() are each declared four times in the file (the second copy immediately above the g_TimeFrame = MapTimeframeInt(TimeFrame) call, and a third+fourth pair interleaved the same way above the second g_TimeFrame assignment), and the g_TimeFrame / g_RSI_Applied_Price globals are declared twice as well. The MQL5 compiler absorbs the duplicate function definitions without complaint, and runtime behaviour is unaffected because each function body is identical, but it is the most visually noisy code-smell in the build. Backtests on H1 XAUUSD with the as-shipped 50-lot default will not reflect the EA's real-world behaviour on a $100 account; reducing LotSize to a value proportional to balance is mandatory before drawing any conclusions from the Strategy Tester report.

The practical operating shape is: load the EA on a single chart, pick a timeframe inside the 1–7 range (H1 is the default and is what the code actually runs when the input is 9), set LotSize to a value the account can support, and let the 14-period RSI do all the work. The 55/82 entry zone is wide enough that signals fire on most days, the 39/35 target keeps each trade short, and the 2-bar timeout is the kill-switch. There is no news awareness, no spread filter, and no session gating, so the EA will continue to attempt entries during illiquid rollover windows and around high-impact releases; if you run it on a live account, either trade through those windows with reduced LotSize, or stop the EA before the calendar and restart it after.