# The Timeseries Refinery

> Open-source platform for versioned (bitemporal) time series storage,
> formula-based computation and dashboarding. Built by Pythonian for
> data-driven teams in energy, trading and industry.

The Timeseries Refinery stores every version of every series: any insert
is kept, and any series can be queried as it was known at any past date
(as-of queries). On top of this versioned store it provides a lisp-like
formula language for computed series with full lineage down to primary
sources, manual-override supervision with audit trail, real-time
dashboarding, an Excel client, a task scheduler, and full Python and
REST APIs. An MCP server lets any MCP-compatible AI assistant query,
compute and analyse the series catalog in natural language.

The Timeseries Refinery is entirely open source (LGPL-3.0-or-later),
written in Python with PostgreSQL storage. The free edition is published
on PyPI, with a delay after the subscriber releases; the pro components
are open source as well, delivered with a subscription. It runs as SaaS (EU-hosted) or self-hosted, with two
major releases per year.

## Documentation

- [API documentation](https://refinery.docs.pythonian.fr/): complete reference for the Python and REST APIs

## Source code

The free-edition components are developed in the open on Heptapod (Mercurial):

- [timeseries-refinery group on Heptapod](https://heptapod.host/pythonian/timeseries-refinery): development sources of the free-edition components
- [tshistory](https://heptapod.host/pythonian/timeseries-refinery/tshistory): core versioned time series storage
- [tshistory_formula](https://heptapod.host/pythonian/timeseries-refinery/tshistory_formula): formula language for computed series
- [tshistory_supervision](https://heptapod.host/pythonian/timeseries-refinery/tshistory_supervision): manual override tracking and audit
- [tshistory_refinery](https://heptapod.host/pythonian/timeseries-refinery/tshistory_refinery): the integrated refinery application
- [tsview](https://heptapod.host/pythonian/timeseries-refinery/tsview): web-based visualisation
- [rework](https://heptapod.host/pythonian/timeseries-refinery/rework): task queue and scheduler

The pro components, delivered with a subscription:

- tshistory_dashboard: advanced dashboarding for complex plots and tables
- tswatch: series update frequency monitoring with alerting
- refinery_pro: OAuth2/OpenID Connect authentication and the MCP server

## Installation

- [tshistory_refinery on PyPI](https://pypi.org/project/tshistory_refinery/): the free edition,
  installable with `pip install tshistory_refinery` — its version number is the version
  of the whole free bundle

## Product

- [Formulas](https://timeseries.pythonian.fr/formulas.html): on-the-fly computation with full traceability
- [Dashboarding](https://timeseries.pythonian.fr/dashboard.html): real-time dashboards over the
  series catalog, particularly well suited to energy commodities (power, gas, CO₂) analytics
- [MCP server](https://timeseries.pythonian.fr/mcp.html): AI assistant integration
- [Comparison with other tools](https://timeseries.pythonian.fr/compare.html): how it differs from other time series databases
- [Blog](https://timeseries.pythonian.fr/blog.html): energy market analyses built with the platform

## Company

- [Pythonian on LinkedIn](https://www.linkedin.com/company/pythonian/)
- [Request a demo](https://meetings-eu1.hubspot.com/marc-schicks)