Pipsgrowth EX15014 SMC-OrderBlock

Pipsgrowth EX15014 SMC-OrderBlock is a liquidity-trap scalper built around a single thesis: when price re-enters a swing-based liquidity zone, the trapped orders on the wrong side create the energy for a fast mean-reversion. The EA doesn't predict — it watches the chart, marks the zones where stops have stacked up, and waits for a confirming surge to enter. The candidate pair and timeframe are XAUUSD on M5, and the EA is structured to react to one setup per bar at most.

The foundation of the strategy is the swing-detection engine. On every new bar of the trading timeframe, IdentifySwingPoints() calls FindSwingHigh() and FindSwingLow() against the high/low arrays using a SwingBarCount = 1 default — meaning a swing is defined as a bar that has a strictly higher high than both of its neighbors (or strictly lower low for swing lows). When a new pivot is found and it isn't a repeat of the previously recorded pivot, AddLiquidityZone() stamps the price into the upperLiquidityZone[] or lowerLiquidityZone[] array, captures the bar time, and calls MarkLiquidityZone() to draw a dashed horizontal line on the chart (crimson for upper, dodger-blue for lower) using OBJ_HLINE and STYLE_DASH. A 0.3% de-duplication threshold (MathAbs(price - zonePrice) / price < 0.003) prevents the same pivot from being added twice if price is just hovering. Each zone is born active, ages out at MaxZoneAge = 24 hours, and is CleanupInactiveZones() once every four hours.

Once zones exist, the EA evaluates them on the next bar. A buy requires the ask to fall inside a lower zone — specifically inside the band [zoneLower, zoneLower * 1.01], a 1% upward buffer above the swing low. A sell requires the bid to fall inside an upper zone — [zoneUpper * 0.99, zoneUpper], a 1% downward buffer below the swing high. After a zone is traded, it is deactivated (zoneActive[i] = false), which prevents the EA from re-entering the same trap on the same test.

The entry decision runs through a six-condition gate inside CheckBuySignals() and CheckSellSignals(). (1) CheckPriceSurge() measures two things — the absolute open-to-current percent change over the last 10 bars and the high-low range percent over the last 5 bars — and uses the larger of the two. The threshold is PriceSurgePercent * 100, which at the default 0.001 translates to 0.1%. This is a very loose floor; on a 5-minute gold bar the open-to-current drift alone almost always clears it, so surge acts more as a sanity check than a filter. (2) IsVolumeSurge() computes a 20-bar average of tick volume, and triggers when the current bar is at least VolumeSurgeMultiplier = 1.5 times that average — but only if RequireVolumeSurge = true, and that input defaults to false, so volume is bypassed by default. (3) IsZoneConfirmedByHigherTimeframe() consults DetermineTrend() on the H4 default (ConfirmationTimeframe = 6 → PERIOD_H4), which returns -1/0/+1 from the sign of MA20 vs MA50; buys need trend = +1, sells need trend = -1, and trend = 0 permissively allows the trade. This gate is also bypassed by default because UseHigherTimeframeConfirmation = false. (4) IsBullishEngulfingPattern() / IsBearishEngulfingPattern() walk back up to EngulfingLookback = 3 bars and look for a current-bar body that fully engulfs the prior opposite-color body's range — current open ≤ prior close and current close ≥ prior open for bullish, mirrored for bearish. (5) IsBullishPinBar() / IsBearishPinBar() walk back up to PinBarLookback = 3 bars and require a body/total-range ratio of PinBarRatio = 0.3 or smaller, a long shadow of at least 2× the body in the rejection direction, and a short opposite shadow of at most 0.5× the body. (6) IsRSIOversold() / IsRSIOverbought() read a 14-period RSI on the trading timeframe and return true when the value is ≤ 30 (oversold) for buys or ≥ 70 (overbought) for sells.

Three of these six gates are wired on by default — engulfing (UseEngulfingPattern = true), pin bar (UsePinBarConfirmation = true), and RSI (UseRSIConfirmation = true). The trend and volume gates are opt-in. The implication is clear: the EA's out-of-the-box behavior is to look for a sweep of a liquidity zone that is also confirmed by a reversal candle pattern and an oscillator extreme. The surge gate is the only always-on check, and it is so loose that it almost never blocks.

The stop-loss and take-profit math is mechanical. For a buy at the ask, stopLoss = zoneLower * (1 - 0.005) — half a percent below the lower zone — and takeProfit = currentPrice + (currentPrice - stopLoss) * RewardToRiskRatio. With RewardToRiskRatio = 2.0 default, the resulting TP is exactly a 1:2 risk-to-reward against the zone-edge stop. The mirror logic applies for sells: stopLoss = zoneUpper * (1 + 0.005), TP at 2× the risk distance below entry. Both levels are passed to Trade.Buy() / Trade.Sell() as broker-side SL and TP, so once the order is filled, position management is the broker's job until the trailing stop kicks in.

ApplyTrailingStop() is wired into the per-tick ManageOpenPositions() loop but is opt-in: EnableTrailingStop = false by default. When enabled, it calculates currentProfit = bid - openPrice for buys (mirror for sells), returns early if the position is not in profit, then sets trailLevel = currentPrice - (currentProfit * TrailingStopPercent / 100.0). At TrailingStopPercent = 50.0 default, the stop sits halfway back from the current price, which means the trade locks in roughly half the open profit once the move has started. The new stop is only applied if it's tighter than the existing one — a one-way ratchet that never loosens — and the modification goes through TryModify_EX15014(), a 3-attempt retry helper that re-issues the modify on TRADE_RETCODE_REQUOTE, TRADE_RETCODE_TIMEOUT, TRADE_RETCODE_PRICE_OFF, and TRADE_RETCODE_PRICE_CHANGED with a 100ms backoff. The same retry pattern is implemented in TryClose_EX15014() and TryClosePartial_EX15014() for completeness, but they are not wired into any live close path in the source — the dead-input caveat applies to those two helpers.

Sizing is honest. CalculatePositionSize() ignores the stopLossPoints argument and returns the fixed FixedLotSize = 0.01 default, clamped to the symbol's min/max/step via NormalizeLotSize(). There is no risk-percentage math, no balance-aware scaling, no Kelly fraction. The EA assumes the operator scales the lot manually. For a $100 account, 0.01 lot is 1% notional at 1:100 leverage and 0.1% at 1:1000 — well within any reasonable broker's comfort zone.

Risk controls are layered. MaxSpreadPoints = 30 rejects any tick whose spread exceeds 30 points — the EA returns early in OnTick() without even calling AnalyzeMarketState(). MaxPositions = 5 caps the open-book size. MaxConsecutiveLosses = 3 triggers pauseUntilTime = TimeCurrent() + PauseMinutes * 60 (60 minutes) inside UpdateTradeStatistics(), after which OnTick() short-circuits and just refreshes the dashboard. EnableTimeFilter = false by default, but if enabled, IsWithinTradingHours() parses TradingHoursStart = "08:00" and TradingHoursEnd = "20:00" and blocks signals outside that broker-time window. AvoidHighImpactNews = true is the most misleading input in the file: the IsHighImpactNewsTime() function does not call any news API. Instead, on first invocation it sets a single hard-coded datetime in a static array (dt.hour = 14; dt.min = 0; ... highImpactNewsTimes[0] = StructToTime(dt)) — a fake 2 PM news event for the current day — and tests proximity to that single timestamp using the 30-minute before/after buffers. The function never refreshes, never fetches, and never deletes the entry, so it's effectively a no-op in real use. Treat this input as a placeholder.

The 39 functions in the file include AnalyzeMarketState() (a 20-bar high-low range volatility print), DetermineTrend() with a 5-attempt progressive-delay retry loop (Sleep(100 * attempts) between failures, max 5 attempts, then invalidates the handle so the next call recreates it), UpdateTradeStatistics() for win/loss/drawdown bookkeeping, and UpdateDashboard() which paints 11 chart labels covering status, volatility, lot+RR, zone counts, trades, profit factor (color-coded green ≥ 2.0, yellow ≥ 1.0, red below), current signal, price+spread, swing distance, and trailing-stop state. On OnDeinit(), all five indicator handles are released and ObjectsDeleteAll(0, "EA_") removes the dashboard and the zone lines in one sweep.

The MapTimeframeInt() helper is defined five separate times in the source — lines 34, 57, 81, 112, and 139 — each with the identical switch statement. The compiler accepts this (later definitions override earlier ones) but it's a code-quality issue rather than a logic bug, and the resulting behavior is correct. If you fork the file, dedupe the helper.

What to expect on a chart: the EA will draw dashed crimson lines above recent swing highs and dashed dodger-blue lines below recent swing lows. When price returns to one of those levels and prints a reversal candle, you'll see a marker line, a buy or sell fires at the next bar open, and a broker-side SL/TP locks in the 1:2 reward. With the trailing stop off, every position is binary — either it hits TP or it hits SL. With the trailing stop on, profitable trades that move in your favor will see the stop ratchet halfway back, locking in roughly half the open profit. Backtests on M5 XAUUSD typically show the strategy working in range-bound hours (London–NY overlap is the most reliable) and struggling in fast directional news spikes. Set MaxConsecutiveLosses = 3, MaxPositions = 5, and leave UseHigherTimeframeConfirmation = false for the default behavior; the news filter, despite its true default, does not actually do anything without a real calendar feed.