MT5 (MQL5) EA — Port my Python DC-WAD Donchian strategy (tick-driven, state machine, partial TP, PSAR/Stoch trailing, optional retracement)

I have a working Python backtester for my “DC-WAD Donchian” strategy. I need a MetaTrader 5 Expert Advisor (MQL5) for live trading that matches the Python logic as closely as possible (no lookahead).

✅ Critical requirement (must accept)

EA must be tick-driven for entries/exits (touch logic).
Bar-close approximation is not acceptable.

Timeframes

• Strategy runs on a single Setup Timeframe (HTF) (user input, e.g., M30/H1/H4 etc.)

• Optional LTF for intraday ambiguity exists in Python, but NOT required for this milestone (we will skip ambiguity mode).

Core logic (must replicate)

1) Donchian + state machine

• Long/short mirror

• States: 00 → 01 → 02 → 03

• Setup01 validity: t+3 is last chance; when count reaches 4 → invalidate/reset

2) WAD filter

• Compute WAD internally + WAD SMA(n)

• Optional HTF WAD filter + optional WAD source mode (as per Python)

3) ATR risk filter

• Reject if risk distance > k * ATR(n)

• ATR must match Python: ATR = SMA(TrueRange, n) (not Wilder)

Entry modes (inputs)

1. Normal (breakout) entry

2. Optional retracement limit entry

• limit at p% retracement (e.g., 50%)

• include time window inputs if needed (signal/fill windows)

Required retracement rule:

• If retracement limit becomes worse than breakout entry:

  • LONG: if limit > breakout_entry → clamp to breakout_entry

  • SHORT: if limit < breakout_entry → clamp to breakout_entry

Trade management

• TP1 closes 50%

• TP2 manages remaining 50% via trailing (input select):

  • PSAR trailing OR Stochastic trailing

• Trailing is tighten-only

• Trailing starts after TP1 as per Python rules

PSAR freeze rule (must implement):

• Maintain last_valid_psar

• LONG: update only when PSAR below close; else freeze

• SHORT: update only when PSAR above close; else freeze

• Use closed bars only (no lookahead).

Execution rules (tick-based)

• Stop/limit triggers by Bid/Ask touch on ticks

• Partial close logic must work in live + tester

• Respect spread (Bid/Ask). No “mid price” assumptions.

Logging

• Export CSV trade log (minimum):

  • entry time/price, exit(s), SL, TP1/TP2, side, session/tag, reason codes, PnL, R-multiple

Deliverables

• Full .mq5 source, clean + commented

• Inputs grouped clearly

• Short README: how to test + any differences vs Python