How to use the Options Strategy Backtester agent skill

What the skill makes the agent do

An agent skill is a folder of instructions and optional scripts that an AI agent loads when a task matches the skill's description. This one covers option portfolio backtesting. Because real option prices are often unavailable, the engine starts from the underlying asset's daily price series and estimates theoretical option prices with the Black-Scholes model, substituting historical volatility for implied volatility.

On top of those estimated prices, the engine simulates profit and loss, portfolio Greek exposure (delta, gamma, theta, vega) and what happens at expiration. The skill is designed as a research tool for testing ideas, not as a live trading system.

• Hedging strategies: covered call and protective put
• Volatility trading: straddle and strangle
• Spread strategies: iron condor, butterfly and calendar spread
• Option pricing analysis and Greeks sensitivity research

Install the options-strategy skill

Download the skill ZIP from this page and unzip it. For a personal setup, place the unzipped folder in ~/.claude/skills/. To limit it to one project, place it in that project's .claude/skills/ folder instead.

If you use the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.

• Personal install: ~/.claude/skills/
• Project install: .claude/skills/ inside your project
• Claude apps: Settings → Capabilities → Skills, then upload the ZIP

Set up the backtest framework and data

The skill does not run on its own. It drives the Vibe-Trading backtest framework and its options engine, so you need that framework available. You also need Python with numpy and pandas, plus Vibe-Trading's src.quantlib.options and backtest.engines.options_portfolio modules.

The historical underlying prices are sourced through the framework. The example config uses tushare as the source, which typically requires a tushare API token.

• Vibe-Trading backtest framework with the options engine
• Python with numpy and pandas
• Vibe-Trading modules src.quantlib.options and backtest.engines.options_portfolio
• Historical price data for the underlying (example config uses tushare, typically needing an API token)

Configure a backtest

A backtest is controlled by a config.json file. The key setting is engine, which must be "options" so the runner selects the option backtest engine. The options_config block holds the pricing assumptions.

Useful fields from the skill: risk_free_rate defaults to 0.05; iv_source currently supports "historical", meaning 30-day rolling historical volatility computed from underlying closes; contract_multiplier defaults to 1.0; same_day_fill defaults to false, which fills a signal dated T on the next bar's close; default_iv defaults to 0.3 and is used for bars without a full rolling window.

• engine must be "options"
• risk_free_rate: default 0.05
• iv_source: "historical" is the supported option
• contract_multiplier: set this correctly for your market
• same_day_fill: false is the default and avoids same-bar fills
• default_iv: default 0.3, must be finite and positive

Write or reuse a strategy engine

A strategy is a signal engine: a class named SignalEngine with a generate method. It receives a map of code to a DataFrame with open, high, low, close and volume columns, and returns a list of instructions. Each instruction gives a date, an action (open or close), the underlying code and a list of legs. Each leg has a type (call or put), a strike, an expiry date and a quantity, where positive is long and negative is short.

You do not have to write one from scratch. The file example_signal_engine.py includes ready-made engines for covered call, cash-secured put, iron condor, vol-gated short strangle and protective put.

• Each leg: type, strike, expiry, qty (negative = short)
• Ready-made engines: covered call, cash-secured put, iron condor, vol-gated short strangle, protective put

Read the output files

After a backtest, files are written to the artifacts/ directory. equity.csv holds daily equity, cash and market value of holdings. metrics.csv holds return, Sharpe ratio, maximum drawdown and similar figures. trades.csv records each open, close, exercise and expire event. greeks.csv holds daily portfolio delta, gamma, theta and vega. ohlcv_{code}.csv holds the raw underlying candlestick data.

• equity.csv: daily equity, cash, holdings value
• metrics.csv: return, Sharpe, max drawdown and similar
• trades.csv: open, close, exercise and expire records
• greeks.csv: daily delta, gamma, theta, vega
• ohlcv_{code}.csv: raw underlying data

Know the limits before trusting results

The Black-Scholes model assumes constant volatility, but real markets show a volatility smile or skew, so pricing can be biased for deep out-of-the-money or deep in-the-money options. The engine models European exercise only, so American-style options or options with significant early-exercise value will be priced incorrectly.

Synthetic mode has no bid-ask spreads or liquidity limits. Real deep out-of-the-money options are often illiquid with wide spreads, so backtest results can be materially optimistic. Contract multipliers also differ by market, so check contract_multiplier before reading any profit figure.

• Constant-volatility assumption biases extreme strikes
• European exercise only
• No bid-ask spreads or liquidity in synthetic mode
• Contract multiplier must match your market

Safety notes: scripts and network access

This skill runs scripts (example_signal_engine.py) and uses the network, mainly to fetch historical price data through the configured source. Review the script before running it, and check which data source and credentials your config uses. The skill is for research and does not place real option orders.

The agent can also call an options_pricing tool for one-off pricing, given inputs such as spot, strike and expiry_days.

• Review example_signal_engine.py before running it
• Network is used to fetch price data; credentials such as a tushare token stay in your environment
• Not a live trading or order-placement tool