Skip to content

Getting Started

This section provides a high-level introduction to how Options Chain Features (OCF) is structured and how it is intended to be used.

Installation

Install OCF via pip:

pip install options-chain-features

OCF has a small runtime dependency footprint and is designed to integrate cleanly into existing Python environments.

Conceptual Overview

At a high level, OCF transforms raw options data through the following stages:

  1. Normalization: Raw vendor-style inputs are mapped into canonical schemas.

  2. Alignment: Underlying data (prices, rates, volatility inputs) are aligned onto a consistent daily calendar.

  3. Feature Engineering: Deterministic feature blocks derive model-ready numeric features.

  4. Greeks & Exposures: Per-option Greeks and aggregate exposure features are computed when option chains are present.

Each stage is explicit, testable and independently reusable.

Canonical Data Model

OCF operates on a small number of canonical tables, the most important being:

  • canonical_underlying_daily: A daily table containing aligned prices, volumes, volatility inputs, and rates.

  • option_chain_snapshot: A snapshot of the option chain at a specific valuation date.

All feature builders and Greeks computations assume these canonical schemas.

Detailed schema definitions are documented in the Schemas section.

Feature Blocks

Features in OCF are grouped into independent blocks, each responsible for a specific class of signals:

  • OHLCV-based price and volatility features
  • Implied volatility level, rank, and dynamics
  • Skew and smile geometry
  • Term structure relationships
  • Liquidity and positioning metrics

Each block operates deterministically, requires explicit input columns and can be enabled or disabled independently. See the Features section for detailed definitions.

Greeks and Exposure Features

When option chain data is available, OCF can compute:

  • Per-option Black–Scholes Greeks
  • Chain-level exposure aggregates
  • Distributional exposure features

These are designed to be model-consistent, free of execution or hedging assumptions and usable independently of the rest of the pipeline. Full details are available in the Greeks section.

Pipelines vs. Building Blocks

OCF supports two usage patterns:

Pipeline-Oriented Usage

Use the end-to-end pipeline to:

  • Normalize raw data
  • Align underlying inputs
  • Build feature tables
  • Compute Greeks and exposures

This is ideal for reproducible research and production workflows.

Component-Level Usage

Import and use individual components directly:

  • Feature blocks
  • Greeks calculators
  • Exposure aggregators

This is ideal for exploratory research and custom workflows.

Both approaches share the same underlying logic and guarantees.

What OCF Does Not Do

OCF intentionally avoids:

  • Data downloading or vendor integrations
  • Trading signals or strategy logic
  • Forecasting or optimization
  • Execution, hedging, or portfolio management

These decisions are left entirely to downstream systems.