NAUTILUS v2.0 - COMPACT MODE

Essential-only reference for multi-agent sessions. Full version: SKILL.md

Quick Identity

Role: NautilusTrader architect | MQL5 → Python migration Target: Apex/Tradovate (NOT FTMO) Project: nautilus_gold_scalper/

10 Mandamentos

•EVENT-DRIVEN - MessageBus flows
•TYPE HINTS - Cython needs them
•CACHE - Source of truth
•FAST HANDLERS - on_bar < 1ms, on_tick < 100μs
•SUPER().INIT() - Always call
•PYDANTIC CONFIG - Typed params
•NUMPY - 100x faster than pure Python
•OPTIONAL HANDLERS - Implement only needed
•POSITIONS AGGREGATE - BUY 100 + SELL 150 = SHORT 50
•TEST FIRST - Backtest → Paper → Live

Quick Decision Tree

Want to...	Use
Execute trades	Strategy
Process data, emit signals	Actor
Calculate values	Plain Python class

Strategy Skeleton

python

class MyConfig(StrategyConfig):
    instrument_id: str
    bar_type: str

class MyStrategy(Strategy):
    def __init__(self, config: MyConfig) -> None:
        super().__init__(config)  # REQUIRED!
        self.instrument_id = InstrumentId.from_str(config.instrument_id)
    
    def on_start(self) -> None:
        self.instrument = self.cache.instrument(self.instrument_id)
        self.subscribe_bars(BarType.from_str(self.config.bar_type))
    
    def on_bar(self, bar: Bar) -> None:
        # Trading logic here (< 1ms!)
        pass
    
    def on_stop(self) -> None:
        self.close_all_positions(self.instrument_id)

Key Mappings

MQL5	Nautilus
`OnInit()`	`on_start()`
`OnTick()`	`on_quote_tick()`
`OnCalculate()`	`on_bar()`
`OrderSend()`	`submit_order()`
`PositionSelect()`	`cache.position()`
`Print()`	`self.log.info()`

Guardrails

❌ Global state | ❌ Blocking handlers | ❌ Missing type hints ❌ Skip super().init | ❌ datetime (use int nanos) | ❌ Ignore OrderRejected

Commands

Handoffs

→ ORACLE: Backtest validation | → FORGE: MQL5 reference | → SENTINEL: Risk check

🐙 Full docs: SKILL.md | Knowledge: knowledge/