Pipsgrowth EX15007 SMC-OrderBlock

EX15007 ICT3 is a low-friction SMC-style scalper that runs a four-detector vote and only takes a trade when the total score clears a configurable threshold. The signal block in IctSignal() walks three closed bars and a SWING_LOOKBACK=12-bar swing window, then assigns weighted points to four independent ICT conditions: 30 points for a Break Of Structure (the previous bar's close pushes through the highest high or lowest low of the swing window), 25 points for a liquidity sweep (the previous bar's wick pierces a swing extreme and the close returns inside), 20 points for a Fair Value Gap (bar[2]'s low sits above bar[3]'s high for bullish, mirror for bearish), and 25 points for an Order Block tap (bar[2] is an opposite-color candle whose body contains bar[1]'s close). Bullish and bearish scores are accumulated independently, and the higher side wins — but only if it exceeds InpMinScore (default 50). The combined ceiling is 100, so a default threshold of 50 means any two of the four detectors firing will normally take the trade, while a stricter setting around 75 forces BOS+FVG+OB to align before the EA acts.

The EA never enters on the signal alone. ComputeRegime() classifies the current bar into one of seven states — StrongTrend, WeakTrend, Range, Breakout, Compress, Expand, Choppy — using a 14-period ADX, the 100-bar ATR percentile, and the live Bollinger Band width at 20/2.0 deviations. The classifier uses two ADX breakpoints: at 1.4× the InpAdxTrendMin input (default 30.8) it splits trend strength, below InpAdxTrendMin*1.4 with ATR-pctile ≥ 80 it marks Breakout, ATR-pctile ≤ 20 marks Compress, and ADX<15 hard-fails to Choppy. The Choppy state is the only regime that NoTradeBlock() blocks outright — Range, WeakTrend, StrongTrend, Breakout, Compress, and Expand all remain tradable, which gives the EA a meaningfully wider operating envelope than EX15 cousins that gate harder on regime.

The confirm layer adds three more filters before TryEntry() will fire. ConfirmEntry() requires that bar[1]'s close agrees with the H1 EMA(50) when InpRequireHTF is on (default true), rejects longs with RSI(14) below 45 and shorts above 55 (a permissive band designed to let trend entries through), and refuses the trade if the current bar's tick volume is below 80% of the 20-bar average. The combination is the only point at which the EA will reject a valid regime+signal combination — the score is the discriminator, the regime is the gate, the confirm is the safety net.

EX15007 is a killzone-time trader. NoTradeBlock() calls ParseWindow() on the two inputs InpLondonWindow (default "08:00-10:00") and InpNewYorkWindow (default "13:30-16:00") and refuses to trade outside both windows, including the entire weekend and the Friday session after 21:00 server time. This is narrower than the EX15 family's typical 7:00-21:00 broad band and matches the real ICT killzone definitions more closely. Combined with the 30-point SYMBOL_SPREAD filter (configurable via InpMaxSpreadPoints), the session gate is what makes the EA's trade frequency match the user's expectation of "two real windows a day, not constant probing."

Risk is keyed off an effective-capital pool rather than raw equity. EffectiveCapital() returns MathMin(InpCapitalCapAmount, equity) when InpCapitalCapAmount>0, else raw equity — so a trader with a $25,000 account and a $5,000 cap trades as if the account is $5,000. CalcLot() then takes 0.5% of that effective capital as the dollar risk per trade and converts it into a lot size using the broker's tick value and tick size, floored to the symbol's lot step and capped at min/max lot. InpCapitalCapFloor=50 refuses entries when the effective capital falls below $50, InpDailyLossLimitPct=3 / InpWeeklyLossLimitPct=9 hard-stop after a daily or weekly loss bracket, and InpCooldownLossCount=3 pauses entries once the EA has seen three consecutive losses on the magic+symbol. Together these four knobs let a user size the EA to a sub-account or a $5,000 paper allocation without rewriting the lot logic.

Stop and target are ATR-anchored with a built-in risk-reward floor. slDist is ATR(14) × 1.5 (InpATR_SL_Mult), capped at MAX_SL_POINTS=5000 points to prevent runaway distances, and the take-profit is ATR × 2.5 (InpATR_TP_Mult), giving a baseline 1.67:1 RR. TryEntry() refuses to send any order with RR < MIN_RR=1.5 after ClampSlToStops() and ClampTpToStops() push the levels away from the broker's stops-level minimum. ManageExits() then runs three layered management passes every tick on every position: (1) a 50% partial close at +1R (PARTIAL_R_MULT=1.0, PARTIAL_PCT=50, only when the partial is at least 2× the symbol's minimum lot), (2) a one-shot break-even at +1R (BE_R_MULT=1.0, the stop ratchets to entry plus the broker's stops level), and (3) an ATR trailing at 1.5× ATR behind price that keeps updating as long as InpEnableTrailing is on. Two exit overrides run on every position check: a 120-bar time exit (MAX_BARS_IN_TRADE) and an opposite-signal exit that calls IctSignal() again and closes if the score flips direction.

The retry paths are wired correctly into the manage loop. TryClose_EX15007, TryClosePartial_EX15007, and TryModify_EX15007 all run a three-attempt loop with 200ms / 100ms Sleep() on REQUOTE / TIMEOUT / PRICE_OFF / PRICE_CHANGED retcodes — the same pattern the EX15 family uses, with the only difference being the helper names. The execution path in SendOrder() also has its own two-attempt retry on the same retcodes and pre-checks OrderCalcMargin + free margin before sending. Together these prevent a price-during-modify race from leaving a position unprotected during the manage pass.

There is one optional scaling layer the trader can switch on. TryPyramid() is gated by InpEnablePyramid (default false), and when on it adds positions in the same direction as the current trade after PYRAMID_ATR_MULT=1.5× ATR of price progress, up to PYRAMID_MAX_LEVELS=3 total positions, never exceeding InpMaxSymbolTrades=2. The pyramid entries reuse the same signal direction and the same 1.5 ATR SL / 2.5 ATR TP scaffold, so each pyramid leg carries the same expected RR — a position-stack rather than a martingale. In dry-run mode, every pyramid prints a [DRY-RUN] PYRAMID line without sending.

EX15007 ships in dry-run by default. InpDryRun=true (the input default) causes TryEntry() to log [DRY-RUN] BUY/SELL lines with the regime code, the score, the lots, and the SL/TP, and return without sending. This is intentional: the file is a configurable SCAFFOLD for the trader to tune, and dry-run is the safe way to validate the score threshold, the killzone window, the capital cap, and the spread filter against the user's broker and symbol before letting it place real orders. InpKillSwitch=true forces OnInit to return INIT_FAILED before any indicator handle is created, giving the user a hard panic-stop in a single input. OnTester() returns (net × profit factor) / (1 + equity-DD$) and zeros out under 30 trades, which biases the Strategy Tester toward strategies with real sample size rather than lucky short runs.

There is a real gap between what the source advertises and what is wired. The 12-layer claim in the header includes SCALING, but in default config SCALING is off — the EA trades one position at a time per symbol until InpEnablePyramid is flipped. The other 11 layers (REGIME, SIGNAL, ENTRY, CONFIRM, NO-TRADE, CAPITAL CAP, RISK, SIZING, MANAGE, EXIT, OnTester) are all live in default config. Traders who want the SCALING layer should turn InpEnablePyramid on and test it in dry-run for a full London+NewYork week before going live, since the same direction signal can stack 3 positions and the 3% daily / 9% weekly caps will scale with that exposure.

This is the right kind of EA for a trader who understands ICT order-flow language and wants a configurable scoring engine. EX15007 is not a set-and-forget grid or martingale — it is a single-position (or optional three-level pyramid) ICT execution shell that respects the killzones, scores the signal honestly, and refuses to trade when the regime is unclear. The risk is genuinely low when the inputs are left at their defaults: 0.5% of capped capital, 1.5 ATR stop, 2.5 ATR target, daily 3% loss cap, weekly 9% loss cap, three-loss cooldown. The risk grows linearly when the trader widens MinScore below 50, widens the killzones, raises InpMaxOpenTrades, or turns on the pyramid — those four knobs are the only paths to the EA taking more exposure than the default design intended.