Pipsgrowth EX17021 Volatility

EX17021 is an envelope-trader that wraps a single ATR(14)-driven band around a fast 10-period EMA on close, then lets the trader pick one of three ways to interact with that band: chase the breakout, chase the breakout and exit at the midline, or fade the band back to the midline. The 'Gaussian' name in the brief is a label, not a kernel — the central line is a plain MODE_EMA / PRICE_CLOSE moving average with Length = 10 bars. The envelope is built on top: upperBand = gaussNow + DistanceMultiplier * ATR(14) and lowerBand = gaussNow - DistanceMultiplier * ATR(14), with DistanceMultiplier = 0.85 out of the box, giving a band roughly 1.7×ATR(14) wide.

A second indicator stack is layered on a different timeframe for trend alignment. TrendFilterTF = 6 resolves to PERIOD_H4 by default, and on that frame the EA plots an EMA(50) on close. The two-band stack is evaluated on the EA's main g_Timeframe (default PERIOD_H1, selectable 1–7 in the inputs), but the HTF EMA cross-check is always read on the trend frame. So a buy breakout on the 1-hour band is only valid if H4 close is above its own 50-EMA, and vice versa for sells. This is the only multi-timeframe filter the EA runs; it does not look at HTF ADX, MACD, or anything else.

Three modes share the same indicator stack but interpret the band differently. StrategyMode_Breakout (default) is a continuation mode: the current bar must close outside the band while the two prior bars closed inside, slope (gaussNow vs gaussPrev) must agree with direction, price must be on the right side of the HTF EMA, and the breakout must travel further than MinBreakDistance * ATR(14) (0.3 × ATR default) past the band. The 'strong candle' filter — body ≥ 60% of candle size — gates the entry one more time, so a doji that pokes through the band is rejected. StrategyMode_BreakoutMeanExit uses the same entry conditions but, on every tick, closes the position the moment price re-crosses the central line in the wrong direction. StrategyMode_MeanReversion ignores the breakout construction entirely; it enters on a single bar's band cross (pricePrev outside, price inside) and closes at the midline. The two mean-touching modes both disable the trailing stop and the breakeven move, so the only exit they will see is the midline or the original SL.

Position management is where the EA does the most work. SL is computed by StopLossOffset(atr), which returns the larger of ATR_MultiplierSL * ATR(14) (1.2 × ATR) and HardSL_Points * _Point (120 points). With UseATR_StopLoss = true the floor dominates most of the time; on quiet pairs the 120-point floor may be the binding number. TP defaults to a flat 160 points (UseATR_TakeProfit = false, FixedTP_Points = 160), and the brief's callout of a 1:1.33 RR comes from this asymmetry. If the user flips UseATR_TakeProfit = true, TP moves to ATR_MultiplierTP * ATR(14) (1.5 × ATR).

For open positions in the two breakout modes, three ratchets run on every tick and the order matters. First, the trailing stop updates SL to bid - slOffset (buy) or ask + slOffset (sell) whenever the new level is better than the current SL — this is a continuous ratchet, not a step. Second, the breakeven move fires once price has travelled one full slOffset in profit: SL is set to entry price, locking the trade flat. Third, the dynamic lock profit runs a discrete dollar ladder: every LockProfitEvery_X_Dollars (default 2 dollars) of unrealized profit, the EA re-anchors the SL to entry + (n * step - buffer) / tickValue * tickSize, so a position with $10 floating locks roughly $8-$9 in price behind the entry. The three mechanisms are designed to stack: trailing gives the first bit of protection, breakeven takes over once the trade is solidly green, and the dollar ladder starts adding a few points of locked profit every additional two dollars. The LockMinusBuffer = 1.0 ensures the lock sits just inside the next dollar mark, which prevents the lock from being immediately tested by spread.

The portfolio side is set up for conservative scaling. MaxOpenTrades = 5 is the ceiling, but OneTradeOnly = true defaults to a single-position configuration. The AllowOnlyProfitableAdditions filter (default true) means that, even if OneTradeOnly is flipped to false, the EA will not add a second position until every open position is showing at least MinProfitPerTradeToAdd = $5. The MaxDrawdownPercent = 10.0 is a hard gate: on every tick, the EA checks 100 * (balance - equity) / balance against this cap and refuses to open new trades when breached. This is the only explicit money-management stop in the EA — there is no daily-loss limit and no per-week reset.

The autotune logic in AutoTuneFilters is small but worth understanding. It reads the current spread and infers the symbol: spreads under 3 points imply a forex pair (pipFactor = 0.0001), wider spreads imply gold or a metals contract (pipFactor = 0.01). minATR is then anchored at 3 × pipFactor, so 0.0003 on EURUSD-class pairs and 0.03 on XAUUSD. This makes the EA portable across the brief's stated range (FX majors and metals) without the user having to swap pip values. The default MinATR = 0.03 matches the gold side; FX users will see the auto-filter drop the floor to 0.0003 once the input is left untouched and UseAutoFilter = true is on.

Order execution itself uses a three-retry scaffold. TryClose_EX17021, TryClosePartial_EX17021, and TryModify_EX17021 each loop up to three times, sleeping 100-200 ms on requote/timeout/price-changed responses and breaking the loop on any other return code. Magic number is fixed at 22217021 and trade.SetTypeFillingBySymbol(_Symbol) lets the broker decide the filling policy. There are no iCustom calls — the three indicator handles are plain iMA and iATR — which keeps the on-chart footprint minimal.

The natural fit is a swing-trader on XAUUSD M5-M15 who wants the band to do the work of locating a regime transition and is comfortable leaving the EA in single-position mode for a session. MeanReversion is for sessions with bounded ranges, Breakout for the same EA when range breaks, BreakoutMeanExit sits between them — it enters like a breakout, exits like a reversion. Backtests will be sensitive to Length, DistanceMultiplier, and TrendEMA more than to any other inputs, because the rest of the management is fixed once OneTradeOnly is chosen.