optionlab-strategy

name	optionlab-strategy
description	Build and validate OptionLab run_strategy inputs as Python dictionaries or Inputs objects, interpret Outputs, and plot strategy profit/loss with plot_pl. Use for coding OptionLab options and stock strategies, including prior positions, different expirations, and terminal-price samples.

OptionLab strategy construction

Create runnable Python for the user's strategy using run_strategy(inputs_data). It accepts an Inputs object or a Python dictionary and returns Outputs in both cases. Prefer the user's requested representation. Answer in the user's language.

This skill describes the repository's OptionLab 1.9.0 API, based on the docstrings and validation in optionlab/models.py, engine.py, support.py, plot.py, utils.py, and the notebooks in examples/. If the installed version differs, check its model fields and docstrings before adapting version-sensitive options. The bundled references are self-contained when installed outside this repository.

Build the input

1. Read the API reference for field names, defaults, units, date rules, calculation selection, and output semantics. It covers every Inputs field and each strategy-leg type.
2. Establish the underlying price, annualized volatility and interest rate, evaluation horizon, price domain, and each leg's side, quantity, strike and premium where applicable. Use supplied data; identify illustrative values as examples. Ask for material missing trade data rather than inventing quotes.
3. Choose dates or an explicit day count. Ensure no option expires before the evaluation horizon. For later expirations, OptionLab values the remaining option at the target date using Black-Scholes.
4. Express rates as decimals and quantities as units: 0.20 means 20%; n=100 means 100 options, with no automatic contract multiplier. Translate contracts to units using the actual contract specification.
5. Select calculations for the requested outputs. Use calculations=[] for only the P/L profile and costs, ["pop"] for probability screening, or omit the field for all calculations. Required market fields remain required in every mode.
6. Validate with Inputs.model_validate(payload) or construct Inputs(...), then call run_strategy. Validation of a model alone does not cover every runtime condition, such as price-domain ordering or integer expirations.

Use the examples for a complete dictionary/object pair, calendar spreads, previous positions, puts, explicit day counts, sample arrays, and plot/export code. Adapt the smallest relevant example; do not execute the entire example catalog for an ordinary user request.

Interpret and present

Read the Outputs table in the API reference before reporting results. Probabilities are fractions; monetary outputs are total strategy amounts. Costs are signed cash flows (debit negative, credit positive). Domain extrema are limited to the sampled plotting domain; they do not establish a global maximum loss or profit. Black-Scholes PoP and conditional expectations cover the full terminal-price domain, including tails outside the plotted interval.

Keep full floating-point values in calculations; format to two decimals only for display. Expected profit and loss are conditional means, not an unconditional expected return. Disabled calculations retain defaults, so zero or an empty list can mean "not calculated". Use pytest.approx for numerical example checks.

Plot

from optionlab import plot_pl
import matplotlib.pyplot as plt

# out is the complete Outputs returned by run_strategy.
plt.figure()
plot_pl(out)
plt.show()

plot_pl(out) draws on Matplotlib's current axes, prints a diagram explanation, and returns None; it does not call show() or return a Figure. To save, use plt.savefig(...) before plt.show(). In a headless process select Matplotlib's Agg backend before importing OptionLab or pyplot. Keep out.inputs and out.data intact because plotting needs them.

Check the generated code

When execution is available, run the chosen example in the user's Python environment, verify it returns Outputs, and inspect the requested fields. For plots, verify a nonempty saved image when running headlessly. Report actual execution results; otherwise state that the code was not run. Do not add notebook magics such as %matplotlib inline or %%time to ordinary Python scripts.