Pipsgrowth EX02034 Breakout

Pipsgrowth EX02034 DonchianExpert is a classic turtle-style breakout EA built on a manually-computed Donchian channel. The default channel period is 20 bars on H1 — the same lookback the original 1980s Turtle Trading system used for its short-term breakout, and the same lookback most breakout retests start with. The channel itself is not pulled from a custom indicator: ComputeDonchian() walks the bar buffer from i+1 to period, finds the highest high and the lowest low, and writes the result into two-element arrays dcs_up[] and dcs_low[] (one value for the just-formed bar, one for the bar before it). The Extremes enum gives five ways to define the channel: pure high/low, with open or close blended in, or capped by open. The default HIGH_LOW_CLOSE mode uses the regular high and the lower of low and close — a slight tightening that tends to reduce false breakouts versus raw high/low. The user can pick which timeframe the channel reads from via InpTimeframe (1-9 mapped to M1/M3/M5/M10/M15/M30/H1/H4/D1), letting the EA run an H1 channel on an M5 chart for faster signals or an H4 channel on H1 for slower, larger-range breakouts.

The signal logic in BuySignal() and SellSignal() is one strict test plus up to three optional filters. The core test is close[0] > dcs_up[0] (long) or close[0] < dcs_low[0] (short). The current-bar close is compared against the channel computed from the previous DonchianPeriod bars. With InpSignalOnBarClose=true (the default), the new-bar gate in OnTick() waits for the candle to actually close before evaluating, so the breakout is judged on a confirmed close, not a wick. If you set it false, the EA fires on the first tick that crosses the level — more reactive, but more prone to spike reversals. All three optional filters ship disabled by default, which keeps the EA close to a vanilla Donchian breakout out of the box; you turn them on to tighten entries in choppy regimes.

The three filters are independent. The ADX filter, when enabled, requires adx[0] >= ADX_level (default 20) and that +DI > -DI for longs, -DI > +DI for shorts. The MACD filter requires the main line above the signal line and above zero for longs (mirror for shorts), with default periods 12/26/9. The RSI/speed filter is a four-RSI confluence: four RSIs with periods 7, 12, 18, and 32 are computed on the weighted applied price, and at least two of the four must agree with the trade direction (above the lower trigger for longs, below the upper trigger for shorts). It also requires MathAbs(close[0] - close[Inp_Bars]) / Inp_Bars >= Trend_lev, a bar-speed gate. The four-RSI design avoids the brittleness of a single-period RSI and is a soft proxy for the multi-timeframe momentum alignment the indicator stack is trying to approximate.

Risk management defaults to a 400-point stop and 600-point target — a 1:1.5 risk-reward — measured in raw price points. OpenBuy() subtracts Inp_StopLoss * _Point from the ask for SL and adds Inp_TakeProfit * _Point for TP. The EA carries no ATR-based sizing and no percentage-of-balance sizing; the lot comes from CalcLot() which respects a basic Inp_Lot (0.01 default) or, if Inp_MMode is set to FREEMARGIN or BALANCE, sizes off the account equity or free margin divided by marginInitial * InpLeverage. That leverage input is a sizing knob, not a real leverage request: the EA will compute equity / (margin * InpLeverage) to figure out how many lots to send, so a 1000 leverage setting produces much smaller positions than a 100 setting. The result is clamped to the broker's min/max/step lot constraints via MathFloor(lots/step) * step in TryBuy_EX02034 and TrySell_EX02034. Each order is preceded by an OrderCalcMargin() check; if the required margin exceeds free margin, the order is rejected with a logged message rather than blowing through and triggering a margin call.

Trailing stop is on by default and uses the simple InpTrailingStep (50 points) behind price, but only after the position has moved at least InpTrailingStart (100 points) into profit. The TrailingStop() function only ever ratchets the SL forward — it never widens it — and only ratchets when the new SL is strictly better than the current one. The break-even module is off by default. When you turn it on, ApplyBreakEven() waits for InpBreakEvenTrigger (100 points) of profit, then moves the SL to openPrice + InpBreakEvenOffset * _Point, which puts the stop two points into profit by default. The profit lock module is also off by default and is mutually dependent on break-even: InpProfitLockStep must be strictly less than InpProfitLockIncrement, otherwise OnInit refuses to start the EA. When profit lock is enabled, it ratchets the SL upward in InpProfitLockIncrement (50-point) chunks as the trade moves in your favor, always keeping InpProfitLockStep (20 points) of distance from the most recent profit floor — a stepped trailing that locks incremental gains rather than a smooth trail.

The pyramid gate is the more interesting position-management feature. Default mode is PyramidGate_ProfitAndSL, which means adding to a position is only allowed if every existing position on the symbol is in profit by at least InpProfitGatePoints (10 points) and the SL of each existing position is already secured in profit (above entry for longs, below for shorts). The PyramidGate_Off mode disables the gate entirely, and PyramidGate_ProfitOnly requires profit but ignores whether the SL is locked. Combined with InpMaxOpenPositions=0 (unlimited by default), the EA will scale into a winning breakout as long as existing positions are net positive and trailing has already pulled the SL into profit. That makes it a true breakout trend-following pyramider rather than a single-entry one.

Order execution is wrapped in a 3-retry loop in TryBuy_EX02034 and TrySell_EX02034, sleeping 200ms between attempts on REQUOTE, TIMEOUT, PRICE_OFF, or PRICE_CHANGED retcodes. Other retcodes (insufficient margin, invalid stops, etc.) break out of the loop and return false, logging the failure with the retcode and description. Modify orders go through TryModify_EX02034 with 100ms sleeps and the same 3-retry ceiling. Inp_Deviation=20 sets the trade object's SetDeviationInPoints for the initial order — the maximum slippage the trade object will tolerate on a market order.

Filling mode is auto-detected in OnInit: the EA reads SYMBOL_FILLING_MODE from the symbol and picks FOK if supported, then IOC, falling back to RETURN. This is the right pattern for ECN/STP/Raw-spread brokers where FOK isn't always allowed. The InpMaxSpread=30 points filter blocks entries when the current spread exceeds the threshold, default 30 points (3.0 pips on a 5-digit broker). Setting it to 0 disables the filter. The InpSignalOnBarClose=true default plus the spread filter plus the optional ADX/MACD/RSI confirmation layers make the EA usable on H1, M5, or anything in between, and the channel can be sourced from a different timeframe than the chart via InpTimeframe, which lets you run, for example, an H4 channel on an M15 chart for medium-frequency breakout trading.

The OnTester() function returns a multi-component fitness score: 30× recovery factor + 15× profit factor + 0.5× min(trades, 200) + 10× sharpe + 25× ROI, then penalties — a 50% haircut if maxDD% > 20, another 70% haircut if maxDD% > 30, and a 50% haircut if balanceDD% > 25. The function returns 0 (and discards the run) if trades < 10, profit ≤ 0, or profit factor < 1.0. There are two bonus multipliers: 1.5× if recovery > 3, profit factor > 2, maxDD% < 15, and trades > 50; 1.3× if ROI > 0.5, profit factor > 1.5, and maxDD% < 20. The composite favors runs that are profitable and low-drawdown, with trade count as a soft tiebreaker — useful for the Strategy Tester genetic optimizer when you want the EA to be punished for over-fitting to a few lucky trades.