The official place to start playing the Numerai tournaments.
Need help? Find us on Discord.
Numerai is quickly developing open-source agent skills for you to use in the tournament. You can start architecting your very own AI scientist. For example:
git clone git@github.com:numerai/example-scripts
cd example-scripts && curl -sL https://numer.ai/install-mcp.sh | bash
codex exec --yolo "find the best neural network architecture to predict target ender"
- Earnings-release features with Kev: score SEC releases with a local language model and join the features by ticker and date.
- Time-series features from Unity: apply the sample prompt to the medium feature set, using each ticker's own history for rolling ranks, average differences, weeks since highs, and weeks since changes.
We highly recommend getting started with Agents using the above section. But, if you're looking to kill some time on artisan data science, you can check out our tutorial notebooks here as well: