Built with Astro, hosted on AWS. Domain registered in 2024.
theforecastingcompany.com is built with Astro, hosted on AWS. It measures traffic with PostHog. It sends email through Brevo (Sendinblue) and Amazon SES.
| Technology | Version | First seen |
|---|---|---|
| JavaScript Frameworks & Libraries1 | ||
| – | 9 Nov 2026 | |
| CDN & Hosting2 | ||
| – | 9 Nov 2026 | |
| – | 9 Nov 2026 | |
| Analytics & Tag Management1 | ||
| – | 9 Nov 2026 | |
| Email Marketing & Lead Capture2 | ||
| – | 9 Nov 2026 | |
| Transactional Email12 | ||
| – | 9 Nov 2026 | |
| Fonts & Icon Fonts1 | ||
| – | 9 Nov 2026 | |
Registration from the registry's RDAP record, the certificate the site serves, and its rank in public traffic lists.
The crew hasn’t read this site’s colors yet. We read a palette when we capture a site’s screenshot, busiest sites first. Meanwhile, browse palettes from other sites.
Counted from every stack change we've recorded: of the sites we watched running one of this site's tools, how many went on to add another. A pattern across sites, not a forecast for this one.
smtp.google.com6crvrobp7f6owc4lp7qryaupisycv7hqmylpcopde4enza27s7nq.mx-verification.google.comfortaleza.ns.porkbun.comcuritiba.ns.porkbun.comsalvador.ns.porkbun.commaceio.ns.porkbun.comoneleet-domain-verification-a337b3bd-9af5-49f2-9712-183e11ebe42bv=spf1 include:_spf.google.com ~allgoogle-site-verification=e9YRnFfDK-RNp4Oc-OElHoFhxv8rOdzLCl4OAfG_V5Ylinkedin-site-verification=8eaf9b96-6020-47b2-9761-f47cc224cc84stripe-verification=E6A9D8F28F9A1726B72D9692E5E8A3FFBFC83B5EB1573C3A7F5894B549B35895v=DMARC1; p=none; rua=mailto:hello@theforecastingcompany.com# The Forecasting Company User-agent: * # ai-train is intentionally left undeclared: no preference stated either way. Content-Signal: search=yes, ai-input=yes Allow: / Sitemap: https://www.theforecastingcompany.com/sitemap-index.xml
# The Forecasting Company > A research lab in Paris building systems that understand time: foundation models for time-series forecasting, zero-shot, with calibrated uncertainty. Backed by Y Combinator. ## Products - [Retrocast™](https://app.retrocast.com/): the browser forecasting workbench — bring a time series, get forecasts with uncertainty intervals - [Forecasting API docs](https://docs.retrocast.com/): canonical developer documentation for the Retrocast™ API; OpenAPI spec at https://api.retrocast.com/openapi - [t0-gamma](https://huggingface.co/theforecastingcompany/t0-gamma): our latest open-weights time-series foundation model, hosted on Hugging Face - [t0-beta](https://huggingface.co/theforecastingcompany/t0-beta): the previous t0 open-weights model, hosted on Hugging Face - [t0-alpha](https://huggingface.co/theforecastingcompany/t0-alpha): the first t0 open-weights model, hosted on Hugging Face - [t0 technical report](https://arxiv.org/abs/2609.24559): architecture, training and benchmark results for the t0 family ## For agents - MCP server (no auth): https://api.retrocast.com/mcp/docs — streamable HTTP; tools: search_forecasting_docs, ask_forecasting_expert. Indexes this site and the API docs. - Docs also publish their own [llms.txt](https://docs.retrocast.com/llms.txt) and [llms-full.txt](https://docs.retrocast.com/llms-full.txt) ## Writing - [A forecast should know what a product is](https://www.theforecastingcompany.com/blog/static-covariates/): A new glove product and a new sandal product sell the same in their first week. Only their category says which season is coming. We show why today's forecasting models cannot use that label, and what changes when the model reads it. - [Introducing t0-gamma](https://www.theforecastingcompany.com/blog/t0-gamma/): Our new open-weights forecasting foundation model ranks first among time-series foundation models on GIFT-Eval, fev-bench and TIME. - [Your next forecast is a SQL query](https://www.theforecastingcompany.com/blog/a-forecasting-model-inside-duckdb/): Run t0-beta ONNX FP16 locally in DuckDB: CPU inference, known-future covariates, multiple quantiles, and offline forecasts after setup. - [Teaching t0 what an empty shelf means](https://www.theforecastingcompany.com/blog/forecast-demand-not-sales/): A stockout flag lets t0 see the demand hidden behind empty shelves. We show what that changes for a seasonal promotion and for the orders a store places. - [200,000 Forecasts a Second](https://www.theforecastingcompany.com/blog/200k-forecasts-a-second/): How we optimized t0-alpha to produce 235,000 full-horizon forecasts and 1.87 million historical forecasts a second in the model core. - [Introducing t0-beta](https://www.theforecastingcompany.com/blog/t0-beta/): Our next forecasting foundation model. More accurate forecasts and a major step forward for the t0 family. - [A foundation model in your browser](https://www.theforecastingcompany.com/blog/a-foundation-model-in-your-browser/): How we built a portable, quantized ONNX version of t0-alpha that runs locally through WebAssembly. - [Describe your data, don’t reshape it](https://www.theforecastingcompany.com/blog/describe-your-data/): Most forecasting APIs make you flatten every series into positional arrays before each request. Our beta API inverts that: describe your columns, grains, and hierarchies once, then any forecast or backtest is one query. - [From ARIMA to Foundation Models](https://www.theforecastingcompany.com/blog/from-arima-to-foundation-models/): Part 3 of The Future of Forecasting. Five eras of forecasting in one diagram, why the latest one changes the contract for practitioners, and where t0-alpha — our first public model — sits in it. - [Time Series Are Not Tables](https://www.theforecastingcompany.com/blog/time-series-are-not-tables/): Part 2 of The Future of Forecasting. Open any data science textbook and you'll find time series stored in CSVs and loaded into DataFrames. That framing hides structural differences that cost real money. - [The Automated Forecaster](https://www.theforecastingcompany.com/blog/the-automated-forecaster/): Part 1 of The Future of Forecasting. Why a forecasting foundation model embedded in an agentic harness can do what the best human forecasters do, at the scale of millions of time series. - [Our CEO on Micode’s Underscore_](https://www.theforecastingcompany.com/blog/micode-underscore/): Geoff sits down with Micode on Underscore_ to talk forecasting, foundation models, and what we're building at The Forecasting Company. - [International Symposium on Forecasting (ISF) 2025](https://www.theforecastingcompany.com/blog/isf/): The Forecasting Company at ISF 2025 - [ShopTalk](https://www.theforecastingcompany.com/blog/shoptalk/): The Forecasting Company at ShopTalk Europe 2025 ## Site - [Blog](https://www.theforecastingcompany.com/blog/): research notes and methodology; this is the blog's canonical home - [Company](https://www.theforecastingcompany.com/company/): who we are - [Contact](https://www.theforecastingcompany.com/contact/): how to reach us - [Legal](https://www.theforecastingcompany.com/legal/): legal notices and terms
| – |
| 9 Nov 2026 |
| – |
| 9 Nov 2026 |
| – | 9 Nov 2026 |
| – | 9 Nov 2026 |
| – | 9 Nov 2026 |
| – | 9 Nov 2026 |
| – | 9 Nov 2026 |
| – | 9 Nov 2026 |
| – | 9 Nov 2026 |
| – | 9 Nov 2026 |
| – | 9 Nov 2026 |
| – | 9 Nov 2026 |