chronos-lab 0.1.8

A batteries-included framework for financial time series analysis.

Welcome to Chronos Lab

chronos-lab is a batteries-included framework for financial time series analysis that turns best-in-class open-source tools into a single, coherent workflow.

It combines ArcticDB for time-series storage, Hamilton DAGs for transparent pipelines, scikit-learn for modeling, and matplotlib for publication-quality visualization—so you can ingest data, analyze thousands of symbols in parallel, and turn results into clear, inspectable insights with minimal glue code.

Prototype interactively in Jupyter notebooks. Scale unchanged pipelines to production with AWS S3 and DynamoDB.

The goal isn’t novelty—it’s leverage. chronos-lab makes the tools you already trust work together, cleanly and predictably.

Quick Links

• Getting Started Guide - Installation, running workflows, common patterns
• Configuration - Configure API keys, storage backends, and environment settings
• API Reference - Complete documentation for all functions and classes
• Tutorials - Interactive Jupyter notebooks with visualizations and step-by-step guides
• Changelog - User-visible features and breaking changes by release

Key Features

Unified Market Data Access : Pull OHLCV time series from Yahoo Finance, Intrinio, or ArcticDB through a single, consistent interface — analysis-ready, UTC-normalized, and pandas-native from day one.

Research-Grade Time Series Storage : Store and retrieve large, versioned time series with ArcticDB, optimized for long histories, cross-sectional analysis, and rapid iteration across large universes.

Pre-Built, Reusable Analysis DAGs : Ready-to-use Hamilton DAGs cover common research workflows from ingestion to features, signals, and diagnostics. Use them as-is, adapt them to your research, or treat them as composable building blocks for new ideas.

Reproducible Research Pipelines : DAG-based execution makes dependencies explicit and results rerunnable — so experiments are explainable, comparable, and easy to extend over time.

Parallel Multi-Symbol Analysis : Apply the same research logic across thousands of symbols efficiently, without hand-rolled batching or orchestration code.

Structured Datasets & Metadata : Manage universes, watchlists, security metadata, and intermediate results as explicit datasets (local files or DynamoDB), keeping research inputs auditable and organized.

First-Class Visualization : Integrated matplotlib plotting for transparent, research-grade diagnostics — inspect signals, anomalies, and distributions directly in notebooks.

Notebook-to-Workflow Integration : Run chronos-lab DAGs interactively in Jupyter, or embed them into larger workflows — from scheduled pipelines in Airflow to event-driven architectures on AWS.

Opinionated, Modular Ecosystem : Install only what you need via optional extras (yfinance, intrinio, arcticdb, aws). No reinvention — just tools designed to work together.

License

MIT License - see LICENSE file for details.