autotrader

name

autotrader

description

Autonomous paper trading bot that manages a stock portfolio on Alpaca's paper trading platform. Use this skill when the user asks to: run the daily trading routine, check the portfolio or account status, execute or close trades, evaluate strategy signals, screen stocks, get technical analysis, review trading performance, deploy a new strategy, run portfolio risk analysis, or validate strategy robustness. Trigger on mentions of paper trading, Alpaca, autotrader, buy/sell stocks, portfolio check, market hours, risk limits, strategy signals, or daily routine.

AutoTrader Skill — OpenClaw Execution Guide

You have two capabilities: analysis (via OBaI CLI) and execution (via Python scripts). Analysis is read-only. Execution mutates your portfolio. Never mix them up.

All commands run from skills/autotrader/.

---

Analysis: OBaI CLI

OBaI is a multi-agent AI system for stock market research. Use obai query to ask any financial question — it routes to the right specialist automatically.

obai query "{your question}" --json --session autotrader_{date}

Rules:

• Always use --json for parseable output
• Always use --session autotrader_{date} (e.g., autotrader_2026-03-21) for context across queries
• The response field in JSON output contains the analysis text
• Check guardrail_rejected and error fields for failures
• OBaI is read-only — never ask it to place trades or manage positions

JSON output structure:

{
  "query": "...",
  "response": "Apple Inc (AAPL) is currently trading at...",
  "agents_called": ["market_data", "fundamentals"],
  "elapsed_ms": 2340,
  "guardrail_rejected": false
}

The response field format depends on the query type:

• General queries (analysis, news, screening, technicals, commodity prices): prose text or numbers
• Portfolio risk/allocation queries: structured risk metrics (Sharpe, Sortino, beta, drawdown, VaR) or sector/asset class allocation breakdown
• Options analytics queries: Greeks computation, scenario P&L grids, position risk profiles with breakevens
• Strategy/backtesting queries: multi-section document where the "Final Strategy JSON" section contains a raw JSON object following the backtest-server's strategy schema (name, universe, indicators, entry_rules, exit_rules, position_sizing, risk_management). Extract and parse this JSON when deploying a strategy to memory/strategies/. The engine supports daily and intraday timeframes (5min, 15min, 1hour), shared-capital portfolio mode (daily only), and walk-forward validation for robustness testing. Ask OBaI for valid indicator types, operators, or timeframe options if needed — it knows the full schema.

Exit codes: 0 = success, 1 = guardrail rejection, 3 = infrastructure error.

Health check:

obai status --json

---

Execution: Trading Scripts

All scripts call alpaca-py directly. They output JSON to stdout. Non-zero exit = failure (check stderr).

Check Market Hours

uv run python -m scripts.market_hours

{"is_open": true, "timestamp": "...", "next_open": "...", "next_close": "..."}

Always check this first. If market is closed, skip trading steps.

Get Portfolio (Account + Positions + Risk)

uv run python -m scripts.get_portfolio

{
  "account": {"equity": 98450.0, "buying_power": 23200.0, "cash": 23200.0, "daily_pnl": -150.0, ...},
  "positions": [
    {"symbol": "AAPL", "qty": 25.0, "avg_entry_price": 195.20, "current_price": 205.80, "unrealized_pl": 265.0, "unrealized_pl_pct": 5.42, ...}
  ],
  "position_count": 3,
  "risk": {"daily_trades_used": 2, "daily_trades_limit": 20, "daily_pnl_pct": -0.15, "current_exposure_pct": 76.5, ...}
}

Execute a Trade

# Market order
uv run python -m scripts.execute_trade --symbol AAPL --side buy --qty 10 --order-type market

# Limit order
uv run python -m scripts.execute_trade --symbol AAPL --side buy --qty 10 --order-type limit --limit-price 195.00

{"order_id": "abc-123", "symbol": "AAPL", "side": "buy", "qty": 10.0, "status": "accepted", ...}

Important: For market buy orders on new positions (stocks you don't already hold), always pass --limit-price with the approximate current price. The risk checker needs a price estimate to calculate position size. For existing positions, it uses the current price automatically.

Risk check is automatic. If the order violates risk limits, you get:

{"error": "Risk rejected: Position would be 15.2% of equity (max 10%)", "allowed": false}

Close a Position

uv run python -m scripts.close_position --symbol AAPL

{"order_id": "def-456", "symbol": "AAPL", "side": "sell", "qty": 25.0, "status": "accepted", ...}

---

Strategy Signal Evaluation

You evaluate strategy signals yourself — no script needed. The process:

1. Read strategy rules from memory/strategies/*.json (entry/exit conditions, indicators, universe)
2. Ask OBaI for the current values of the indicators defined in the strategy for its universe symbols — request just the numbers
3. Compare each indicator value against the strategy's entry/exit rule thresholds and operators in your own reasoning
4. For each symbol, determine: entry signal, exit signal, or no signal

Ask OBaI for exactly the indicators your strategy defines — don't request indicators the strategy doesn't use. If you need to know what indicators or operators the engine supports, ask OBaI directly — it can list them.

You can also ask OBaI to analyze your current portfolio's risk or sector concentration at any time — useful before making entry decisions or during the daily journal.

---

Environment Variables Required

ALPACA_API_KEY=your_key       # Required for all trading scripts
ALPACA_SECRET_KEY=your_secret # Required for all trading scripts
OPENAI_API_KEY=sk-proj-...    # Required for OBaI analysis

Optional risk limit overrides (defaults in parentheses):

MAX_POSITION_PCT=10.0         # Max single position as % of equity (10%)
MAX_DAILY_TRADES=20           # Max trades per day (20)
MAX_DAILY_LOSS_PCT=3.0        # Stop trading if daily loss exceeds (3%)
MAX_EXPOSURE_PCT=90.0         # Max invested capital as % of equity (90%)