alpacalyzer-algo-trader: Skill for Claude Code

Scope Constraint

• Strategy files go in src/alpacalyzer/strategies/{name}.py
• Tests go in tests/strategies/test_{name}.py
• Strategies evaluate entry/exit conditions based on signals

Placeholders

• <strategy> — lowercase (e.g., mean_reversion)
• <Strategy> — PascalCase (e.g., MeanReversion)

Steps

1. Study the reference implementation

Read these files in order:

1. src/alpacalyzer/strategies/base.py — Strategy protocol, BaseStrategy, EntryDecision, ExitDecision
2. src/alpacalyzer/strategies/config.py — StrategyConfig dataclass
3. src/alpacalyzer/strategies/momentum.py — canonical implementation

Key concepts: strategies implement evaluate_entry() and evaluate_exit(). They receive TradingSignals + MarketContext and return decision objects. Agent recommendations are optional inputs.

2. Create strategy file

Copy src/alpacalyzer/strategies/momentum.py → src/alpacalyzer/strategies/<strategy>.py and modify:

• Config: create DEFAULT_<STRATEGY>_CONFIG with strategy-specific params
• evaluate_entry() — implement your entry logic. MUST include stop_loss in every EntryDecision(should_enter=True)
• evaluate_exit() — implement exit logic with urgency levels (normal, urgent, immediate)
• Use self._check_basic_filters() for standard guards (market open, cooldown, existing position)

3. Register strategy

Edit src/alpacalyzer/strategies/registry.py — add to _register_builtins().

4. Write tests

Create tests/strategies/test_<strategy>.py following the pattern in tests/strategies/test_momentum.py:

• Test entry with bullish/bearish signals
• Test market closed, existing position, cooldown rejection
• Test position sizing stays within limits
• Test exit for profitable and losing positions
• Test catastrophic drop triggers immediate exit
• Test agent recommendation integration

5. Run and verify

uv run pytest tests/strategies/test_<strategy>.py -vv
uv run pytest tests/strategies/  # regression