Pipsgrowth EX08005 MACD

EX08005 sits at the slow end of the PipsGrowth MACD family. Where most MACD EAs run the textbook 12/26/9 pair on M5 and react to every wiggle, this one runs a 120/260/90 MACD — the longest standard deviation of the slow EMA period across the corpus — and the goal is to trade only the swings that survive a multi-month filter. The default chart is XAUUSD on M5 (H1 also fits), the magic number is 22208005, and the version is 2.00.

The signal is built from four indicator handles created in OnInit. Three of them are moving averages: a 55-period EMA applied to close (the fast MA), a 69-period SMA applied to close (the slow MA), and a 2-period LWMA applied to close (the filter MA). The fourth is the MACD with a 120-period fast EMA, a 260-period slow EMA, and a 90-period signal-line EMA — a 90-bar smoothing of the difference between two very slow EMAs. On each new bar the EA reads MACD_MAIN and MACD_SIGNAL at bar 1 (the most recent closed bar) and at bar 3 (two bars back). The decision rule in OnTick is plain: a long fires when MACD_MAIN[1] is above MACD_SIGNAL[1] AND MACD_MAIN[3] was below MACD_SIGNAL[3]; a short fires on the mirror image. That second condition is the trick — it requires the crossover to have completed two bars ago, which filters out a single-bar whipsaw and forces the EA to wait for the new relationship to hold.

The MAs are computed on every tick but, in this build, do not gate the entry. MAFast_0, MASlow_0, MAFilter_0, and MAFilter_2 are all read into local doubles, but the entry branch only inspects the four MACD values. Anyone tuning the source who wants a true MA filter has to add the comparison themselves; out of the box, the MAs are declared and unused as signal inputs. This is a deliberate simplification — the long-period MACD already encodes a slow trend filter — but it is worth knowing if you expect the header's wording about "MA Fast/Slow/Filter confirmation" to mean those three MAs veto entries.

Position management runs in priority order on every open ticket. First, the EA checks for an opposite MACD signal — if a long position exists and MACD_MAIN[1] has fallen below MACD_SIGNAL[1] while MACD_MAIN[3] was above MACD_SIGNAL[3], the position is closed via TryClose_EX08005 (3 retries, 200ms backoff on REQUOTE / TIMEOUT / PRICE_OFF / PRICE_CHANGED). Second, the EA applies a partial close: when price moves 70 pips in the position's favour (configurable via InpProfitOne), half the lot is closed through TryClosePartial_EX08005. Third, if InpBreakeven is non-zero (default 0 — disabled), the SL is dragged to the open price once price is InpBreakeven pips in profit. The order of those three checks is continue-stacked in OnTick, so each tick performs at most one management action per position. The hardcoded initial SL is 80 pips and the hardcoded initial TP is 500 pips — a 1:6.25 R:R bracket that lets the long-period MACD play out while keeping the per-trade loss shallow.

Money management is a single switch. With MM=false, every entry uses InpLots=1.0. With MM=true, the EA instantiates a CMoneyFixedMargin object and sizes each entry off the SL distance at 1% of free margin (Risk input). In both cases LotCheck() then normalizes the requested lot to the symbol's step / min / max. Filling mode is auto-detected per symbol — FOK first, then IOC, then RETURN — and m_slippage=10 is the maximum deviation.

The hardening stack is the most defensive of the EX08 series. IsSafeToTrade() runs nine independent gates before any entry: a capital-allocation cap (off by default; set InpCapitalCapAmount to the dollar ceiling and InpCapitalCapFloor=50 to the dollar floor), a 95% equity floor relative to the initial balance, a 3% daily realized-loss limit and a 6% weekly realized-loss limit (both computed against effective capital so the cap interacts cleanly with the loss limits), a 3-consecutive-loss counter that triggers a 30-minute cooldown (InpCooldownMinutes), a 50-trade daily cap, a market-hours check that closes entries on Saturday, before Sunday 22 GMT, and after Friday 22 GMT, a London-or-NY session check (London 7–16 GMT, NY 12–21 GMT, Asia 0–7 GMT explicitly avoided), and an optional 15-minute news filter around the London and NY opens. The trade counters and the realized PnL stream update in OnTradeTransaction, which scans deal history for DEAL_ENTRY_OUT events on this magic and this symbol.

GMT handling is the standard PipsGrowth auto-detect: DetectGMTOffset() walks back through the H1 bars until it finds a weekend gap, then computes the broker's offset from the Friday 22 GMT close. InpServerGMTOffset=0 turns auto-detection on; set it to a positive or negative integer to override. The session check, news check, and weekend guard all run on this offset, so the EA's idea of "London open" follows the broker's clock rather than local time.

OnTester returns (STAT_PROFIT × STAT_PROFIT_FACTOR) / (1 + STAT_BALANCE_DDREL_PERCENT) with a hard minimum of 30 trades. A trade count below 30 zeroes the result, which pushes the optimiser towards parameter combinations that actually trade instead of sitting in cash. There is no upper bound on the formula — a strategy with 2.0 PF and 8% relative drawdown scores roughly 1.85; the same PF with 25% DD scores roughly 1.60, so the drawdown term has real weight.

In practical terms: this is an XAUUSD MACD crossover EA on M5/H1 with a 120/260/90 slow MACD, a 1:6.25 R:R default bracket, half-position partial close at +70 pips, a nine-gate no-trade stack, and a defence-in-depth risk layer. It will trade rarely — long-period MACD signals on a 5-minute chart cluster only around real regime shifts — and when it trades it cuts losers at 80 pips while letting the partial close lock in the first 70 pips of any move. The expected backtest profile is a low-trade-count equity curve with intermittent bursts when the slow MACD finally turns. Anyone running it should size off the 80-pip SL with the assumption that the next signal might be a week away, and that the session filter will close the book before the daily reset.