Alternative Macro Signals specializes in extracting actionable macroeconomic and commodities signals from unstructured text data.

Our clients include investment firms and central banks.

We identify relevant news and transform them into quantitative signals that either drive investment decisions or help understand where the economy is heading to.

• Daily Inflation Indicators - inflation diffusion indices (NIPI), news volume indices,...

• Inflation NewsBot - News potentially relevant to near-term inflation forecasting

• Commodities Signals - Fundamentally relevant insights from news flow

• Alternative Economic Indicators - Beyond traditional metrics

Our platform combines:

• Web Scraping - Comprehensive data collection, global coverage

• Large Language Models - State-of-the-art models covering multiple languages

• Supervised Learning - Specific tasks taught by economists to the LLMs

• Real-time Processing - Daily frequency data, if not refreshed throughout the day

Our GitHub pages are mostly dedicated to documenting our API and SDK for existing customers to access our data.

• api-docs documents our endpoints and how to use them

• ams-sdk is our API Python wrapper for quick access

For reference, our client portal gathers all the different ways to access our data (including in-app).

Contact support for any client enquiries, including API credentials requests.

If you are less familiar with our data,

• Check our website for comprehensive data description

• Reach out to get an introduction to our datasets and temporary access

• Test our data, including historical time series, for a limited period

• The best results with LLMs are obtained by fine-tuning Language Models, and having a field specialist (an economist) frame the problem and teach the Language Models the very specific knowledge they need

• We have historical data starting 1/2018, with "frozen models" and point-in-time data

• Exceptional global reach

This is the official public GitHub for Alternative Macro Signals.