Pipsgrowth EX15008 SMC-OrderBlock

Pipsgrowth EX15008 is an ICT-stack M5 scalper on XAUUSD that scores the inner-bar structural footprint (BOS + liquidity sweep + FVG + order-block tap) into a single 0-100 confidence value, then gates the result behind an H1 trend filter, a 7-state market-regime classifier, and an explicit London / New-York killzone window. The default InpMinScore = 50 is intentionally moderate: a long signal is admitted as soon as the bull-side score crosses fifty, which in this code means roughly two of the four sub-detectors firing (BOS contributes 30 points, sweep 25, order-block 25, and the FVG gap adds a further 20). The EA is shipped with InpDryRun = true, so a fresh install logs every would-be order to the Strategy Tester journal but never places a ticket; flipping that input to false is the only switch required to go from paper-trading to live.

Inside each new closed bar, OnTick calls IctSignal() first. IctSignal walks a SWING_LOOKBACK = 12 bar window to find the most recent swing high and swing low, then checks four things on the last three closed bars: (a) whether bar[1] closed through that swing (BOS), (b) whether the wick of bar[1] pierced the swing and the body closed back inside (liquidity sweep), (c) whether bar[2]'s low sits above bar[3]'s high, or bar[2]'s high sits below bar[3]'s low (FVG), and (d) whether bar[2] is an opposite-color candle whose open-close range contains the bar[1] close (order-block tap). Each component that fires adds its weight to a running bull- and bear-score, and whichever side clears InpMinScore with the higher total wins. The two attempts never tie because a single bar can only satisfy bullish conditions or bearish conditions for any given detector.

A winning signal is then run through ConfirmEntry(), which reads the H1 EMA(50) handle (g_h_emaH) and rejects a long whose M5 close is below the HTF EMA, a short whose M5 close is above it, then applies a directional RSI band — longs need RSI(14) ≥ 45, shorts need RSI(14) ≤ 55 — and finally checks that the current tick volume exceeds 80 % of the 20-bar average. Any one of these failing returns a rejection string that the EA never sends. The Net effect is a triple-filter entry: a structurally-weighted ICT score, a higher-timeframe trend agreement, and a microstructural vol/RSI band. ComputeRegime() runs the same tick to classify the market as StrongTrend, WeakTrend, Range, Breakout, Compress, Expand, or Choppy using ADX(14) ≥ InpAdxTrendMin = 22 plus a 100-bar ATR-percentile and a Bollinger-width band, and a Choppy state hard-blocks the trade via NoTradeBlock().

Position management is a three-stage lock-up. When a position reaches 1R (one times the initial risk, defined as 1.5 × ATR), the EA fires PARTIAL_PCT = 50 of the position through TryClosePartial_EX15008 — provided the remaining volume is at least twice LotsMin. At the same 1R threshold, InpEnableBreakeven moves the stop to the entry price with TryModify_EX15008. From there, the stop is re-anchored every tick to (current price minus 1.5 × ATR for longs, plus 1.5 × ATR for shorts), a simple fixed-ATR trail rather than the more elaborate step-trails some EAs use. The two safety exits are MAX_BARS_IN_TRADE = 120 (10 hours on M5, 30 hours on M15 — far longer than the 60-bar limit of the EX15 cousins) and an opposite-signal exit: every tick, ManageExits re-runs IctSignal against the same magic, and if the score flips sign the position is closed through TryClose_EX15008. All three manage helpers retry up to three times on REQUOTE, TIMEOUT, PRICE_OFF, or PRICE_CHANGED with a 200 ms / 100 ms Sleep between attempts; SendOrder() additionally has its own 2-attempt inner retry on the same retcodes before it gives up and returns false.

Risk control runs on effective capital rather than raw equity. EffectiveCapital() takes MathMin(InpCapitalCapAmount, g_acc.Equity()) when the cap is set, falling back to raw equity when InpCapitalCapAmount is 0. RiskGateOK() then refuses new entries when today's realized PnL has fallen below –3 % of eff_cap, the week is below –9 %, g_effCap has dropped below InpCapitalCapFloor = 50, the master kill switch is on, or the consecutive-loss run has reached InpCooldownLossCount = 3. Position size itself is calculated from the per-bar ATR(14) via CalcLot: lots = (eff_cap × InpRiskPercent / 100) / (slDistance × moneyPerLotPerPrice), floored to InpFixedLot = 0.05 if the calculation breaks, then rounded down to the broker's lot step and clamped to [LotsMin, LotsMax]. TryEntry additionally rejects an order whose effective R-multiple after stops-level clamping falls below MIN_RR = 1.5 — that check is the runtime safety net against brokers that quote gold with a 50- or 100-point stops level that would otherwise eat the SL.

The no-trade stack is short and explicit. NoTradeBlock() requires the live spread to be below InpMaxSpreadPoints = 30, the day to be a weekday, the clock to be inside either InpLondonWindow = "08:00-10:00" or InpNewYorkWindow = "13:30-16:00" (server time after InpTimeZone is applied), the broker's trade flag to be on, and the symbol's trade mode to be FULL, LONG_ONLY, or SHORT_ONLY. Friday after 21:00 and the full weekend are blocked. There is no news-blackout filter in this build — the developer left that to the trader. A duplicate-entry guard (g_lastEntryBar) prevents the EA from opening a second position on the same bar after the first; InpMaxOpenTrades = 2 and InpMaxSymbolTrades = 2 then cap the live exposure.

The pyramid layer is implemented and wired but defaulted off. When InpEnablePyramid = true, TryPyramid() runs after every entry attempt: it counts existing positions for this magic on this symbol, refuses to add if the count is at PYRAMID_MAX_LEVELS = 3 or InpMaxSymbolTrades, then opens another position only when the current price has moved at least PYRAMID_ATR_MULT = 1.5 × ATR beyond the most recent open price in the signal direction. The new leg uses the same SL = 1.5 × ATR, TP = 2.5 × ATR template and the same CalcLot sizing, so each pyramid layer is sized like a fresh entry rather than scaled by layer. With the default off, the EA never pyramids; turning it on is a one-click change in the inputs but it changes the strategy character from single-shot to scaling, and the tester results should be re-run with it on.

The OnTester custom criterion is (net × profitFactor) / (1 + equityDD), and it returns 0 if the test produced fewer than 30 trades. That single formula drives the Strategy Tester's ranking of every optimization pass, so the EA's optimization surface is implicitly biased toward parameter sets that produce a reasonable trade count with positive expectancy and a low drawdown. The five indicator handles — ATR(14), ADX(14), Bollinger(20, 2, 0), RSI(14), and the HTF EMA(50) on H1 — are released in OnDeinit, and g_effCap is recomputed at the top of every tick so the cap never goes stale during a long session.

For installation: minimum recommended balance is $100, InpRiskPercent = 0.5 % means a $100 account risks roughly $0.50 per trade, and a $1,000 account risks $5. The strategy is built for XAUUSD M5 (and M15), so a low-spread, GMT+2 or GMT+3 broker is required to keep spreads below the 30-point gate; high-spread retail books will see the EA sitting in no-trade for most of the day. Set InpCapitalCapAmount to a value lower than your account balance to test the strategy as if it were trading a sub-account (e.g. cap = 200 on a $1,000 broker account means the EA behaves as if it were running on a $200 account), and keep InpDryRun = true for at least one full London + New York session before going live so the journal confirms that signals are firing on the right bars and that the regime classifier is settling on sensible labels.