Pipsgrowth EX03001 Grid

Pipsgrowth EX03001 Thanos Grid Adaptive is a stop-order grid built for traders who want to layer into a position when price moves against the first entry. The EA is broker-portable and ships with a 12-layer execution stack wired in fixed order: REGIME, SIGNAL, ENTRY, CONFIRM, NO-TRADE, CAPITAL CAP, RISK, SIZING, MANAGE, EXIT, SCALING, OnTester. Magic number 22203001, version 2.00.

How the grid is actually built. The first trade is a market order generated by the signal engine. Once that order is on, ManageGrid() places a pending stop at a distance of InpGridATRSpacing times ATR (default 1.5 times ATR(14)) from the current bid or ask, in the same direction as the signal. The pending is a BUYSTOP above ask on a long or a SELLSTOP below bid on a short. When (and if) price continues against the position and triggers the stop, ManageGrid() queues the next stop, with lot size scaled by InpLotMultiplier raised to the level (default 1.3) - level 1 keeps the base lot, level 2 multiplies by 1.3, level 3 by 1.69, level 4 by 2.20, level 5 by 2.86. The grid is hard-capped at GRID_MAX_LEVELS = 5 per side, so the EA will not place a 6th stop in the same direction regardless of input settings.

Important default-state note. InpPyramidEnabled defaults to false. Out of the box, the EA only takes the initial market trade and never activates the pending-stop layers. To engage the grid, the user must flip InpPyramidEnabled to true. This is deliberate: the EA is shipped with grid scaling OFF so the trader can verify the signal and the position management logic on a single trade before committing to multi-layer scaling. The lot-multiplier input (default 1.3) and the grid spacing input (1.5 times ATR) only matter once the grid is enabled.

Signal formation. The signal engine runs only on a new bar. On the closed bar at shift 1, the EA compares close[1] to the upper and lower Bollinger Band (period 20, deviation 2.0). If close[1] breaks above the upper band AND the regime classifier returns a trend-permitting label, the EA produces a +1 buy signal. A close below the lower band produces a -1 sell signal under the same condition. Confidence is logged at 55 for the band break. A fallback fires when the bar body exceeds 0.7 times ATR - bullish body yields +1, bearish body yields -1, with confidence 40. The fallback exists so the EA still trades after a large-range expansion even if the close stays inside the bands.

Regime classifier. ClassifyRegime() reads ATR(14) and ADX(14) on the current timeframe plus Bollinger Bands (20, 2.0) on the current timeframe. It computes the ATR's 100-bar percentile rank against a 25/75 threshold and the BB-width's 100-bar percentile rank against a 20/80 threshold. Seven labels come out: Breakout (ADX at or above 25 with both ATR and BB width in their top quartile), StrongTrend (ADX at or above 25 outside the breakout condition), WeakTrend (ADX between 18 and 25 outside compression), Range (ADX between 18 and 25 in compression), Expand (ATR or BB width top quartile with ADX below 18), Compress (ATR or BB width bottom quartile with ADX below 18), and Choppy (the default catch-all). The signal engine treats only StrongTrend, WeakTrend, Breakout, and Expand as tradeable. NoTradeCheck() adds a separate hard gate: if InpUseRegimeGate is on, ADX below 15 blocks entries outright.

Confirmation layer. ConfirmEntry() rejects the trade if ATR is zero (data warmup), the session check fails (in live mode), the HTF EMA(50) on H1 disagrees with the direction (price below the EMA on a long, above on a short), RSI(14) is overbought on a long (above 70) or oversold on a short (below 30), or the computed reward-to-risk ratio is below 1.2. The HTF EMA(50) on H1 is the one-sided trend-direction filter - it pulls the grid into the higher-timeframe trend and stops the EA from buying into H1 resistance or selling into H1 support. The 1.2 minimum R/R is enforced as a ratio between the fixed 2.5R TP and the 2R SL distance, so it always passes in normal conditions but fails if ATR drops to zero.

Risk and sizing. Position size is computed by CalcLots(): take InpRiskPercent (0.5% default) of effective capital, divide by the loss-per-lot for the SL distance (InpATR_SL_Mult times ATR = 2 times ATR by default), then clamp to the broker's lot min/max/step via ClampVolume(). Effective capital is the lesser of InpCapitalCapAmount and account equity - so with InpCapitalCapAmount set to 0 (the default), the cap is disabled and full equity is used. InpCapitalCapFloor (default 50) blocks new entries when the capped effective capital drops below $50.

Daily loss limit is 3% of effective capital; weekly loss limit is 6%. Both are computed against deals with this magic number (DEAL_MAGIC equals 22203001) and applied via MagicDayPnL() and MagicWeekPnL() inside NoTradeCheck(). A 30-minute cooldown is triggered after 3 consecutive losses, set in OnTradeTransaction() and honored by NoTradeCheck() until the cooldown timer expires.

The kill switch input (InpKillSwitch) is the emergency close - when set to true, the EA calls CloseAll() on the next tick, cancels pending orders, and refuses new entries until the input is flipped back.

Position management. For every open position, ManageOpenPositions() runs each tick and applies four management rules in this order. First, break-even: when price reaches 1R from entry (where 1R equals InpATR_SL_Mult times ATR), the SL is moved to entry plus the broker's stops_level so the trade cannot go negative. Second, ATR trailing: a Chandelier-style trail at 2.5 times ATR(14) below the current bid (for longs) or above the current ask (for shorts), applied only when the trail is forward of the current SL. Third, time exit: positions held longer than 600 bars on the current timeframe are force-closed. On M5 that is 50 hours, on H1 it is 25 days. Fourth, regime Choppy exit: if the regime classifier returns Choppy on a subsequent bar, the position is closed regardless of profit or loss.

TryModify_EX03001() retries the SL/TP modification up to 3 times at 200ms intervals on REQUOTE, TIMEOUT, PRICE_OFF, or PRICE_CHANGED retcodes. TryClosePartial_EX03001() does the same for partial closes used inside the time-exit and regime-exit paths.

Session handling. InpUseGMTSessions defaults to false. When left off, the EA trades 24/5 on weekdays and the InSession() helper simply blocks Saturday and Sunday. When turned on, InActiveSession_EX03001() requires the GMT hour to be inside either London (7 to 16 GMT) or New York (12 to 21 GMT). DetectGMTOffset_EX03001() runs at init and walks the offset range -12 to +12 to find the score that places the server time inside a weekday 7-21 GMT window - a small auto-calibration step that avoids hardcoding the broker's GMT offset in the source.

OnTester fitness. The tester pass returns (netProfit times profitFactor) divided by (1 plus balanceDrawdown), with a hard gate of 30 or more trades. If the EA trades fewer than 30 times in the backtest, OnTester returns 0 and the genetic optimizer discards the pass. This biases the optimizer toward inputs that produce enough trades to be statistically meaningful.

What to expect in a backtest. Because the entry is on a new-bar closed-candle breakout, the EA naturally skips inside bars and re-enters on the next confirmed breakout. The signal engine produces relatively few trades on its own (a handful per week on M5 XAUUSD), but once InpPyramidEnabled is true, each trade can spawn up to 5 grid layers. With the default 0.5% risk per layer, a fully-filled 5-layer grid commits 0.5% times (1 + 1.3 + 1.69 + 2.20 + 2.86), or about 3.8% of equity, to a single direction. On a $100 account that is only $3.80 of risk in absolute terms, which is why the source declares a $100 minimum deposit - but a serious grid user should size up to $1,000 or more to absorb the worst-case multi-layer drawdown on XAUUSD's typical 1.5 to 2.0 ATR intraday range.

Design asymmetry. The EA is built as a stop-order grid, not a market-execution grid. The 1.5 times ATR pending-stop distance is the key parameter - smaller spacing means more layers fill in normal volatility, larger spacing means fewer layers but tighter risk on a real breakout. Combined with the 2 times ATR hard SL on each layer and the 2.5R final TP, the design is asymmetric: a single 1R move covers the risk on the first layer; a 2.5R move pays for the entire trade plus the unfilled layers' opportunity cost.