tools
Numerai Tournament
A machine-learning tournament where participants predict stock-market returns, submit models, and optionally stake NMR on performance.
In inglese
Numerai Tournament provides an obfuscated financial dataset for building machine-learning models. Participants submit prediction files, receive scores, and build a public track record on the leaderboard. Numerai combines submissions into a Meta Model used by its hedge fund.
Participation and scoring are free without staking. Users may stake NMR on their own predictions to receive rewards or lose part of the stake. The obfuscated data is intended for the tournament and cannot be mapped directly to tradable stocks. Model Uploads and Numerai CLI automate submissions.
Features
- Download a free, obfuscated financial dataset
- Build models using official Python and R example scripts
- Submit prediction files through Numerai's API
- Track CORR and Meta Model Contribution scores
- Stake NMR on predictions for potential rewards or burns
- Automate daily submissions with free Model Uploads
- Run self-hosted automation with Numerai CLI
- View rankings, scores, payouts, and reputation on the leaderboard
Use cases
- Train machine-learning models to predict stock-market returns
- Submit predictions and build a public performance record
- Test model quality without buying market data
- Automate recurring predictions with Model Uploads or CLI
- Stake NMR on models with confidence in their performance
Pros
Cons
Latest updates
- UX for the Modern Era
Replaced Home with Onboarding, unified model pages in a Dashboard, and added model archiving and AI manager mode.
Capabilities
- Spreadsheets — “submission.to_csv(f"submission.csv")” source
- Builds models — “You build machine learning models on it.” source
- API — “from numerapi import NumerAPI” source
- Self-hosted — “Also known as "Compute Heavy", this is an official self-hosted cloud solution that supports AWS, GCP, or Azure.” source
Get it
Pricing
- Prices checked
- 2026-09-26