# Numerai Tournament

> Numerai is a free, recurring data science competition where machine learning models predict the global stock market. Numerai provides a stock market dataset, evaluates submitted predictions, and combines staked models into the Meta Model used by the Numerai hedge fund.

## How the tournament works

- Numerai publishes free tabular training, validation, and live data. Each row represents a stock at a point in time, with features and targets for modeling.
- Participants train models to predict relative future stock returns, then submit ranked predictions for each live round.
- Submissions are evaluated over time. Core metrics include Correlation (CORR), which measures performance against the target, and Meta Model Contribution (MMC), which measures a model's additive value to the ensemble.
- Participation and scoring are free. Participants may optionally stake Numeraire (NMR) on a model; positive performance can earn NMR and negative performance can burn part of the stake.
- Only staked predictions are combined into Numerai's stake-weighted Meta Model, which informs the Numerai hedge fund.

## Start building

- [Getting Started](https://docs.numer.ai/): A concise introduction to the dataset, modeling, submissions, scoring, and staking.
- [Tournament Data](https://docs.numer.ai/numerai-tournament/data): Dataset structure, features and targets, versions, and downloads.
- [Models](https://docs.numer.ai/numerai-tournament/models): Tutorials, model slots, benchmark models, validation, and ensembling.
- [Submissions](https://docs.numer.ai/numerai-tournament/submissions): Live rounds, prediction formats, validation, and upload workflows.
- [Scoring](https://docs.numer.ai/numerai-tournament/scoring): CORR, MMC, Feature Neutral Correlation (FNC), score calculation, and seasons.
- [Staking](https://docs.numer.ai/numerai-tournament/staking): Optional NMR staking, payouts, burns, and risk.
- [API and MCP](https://docs.numer.ai/numerai-tournament/api): Official GraphQL API, NumerAPI Python client, and Model Context Protocol access.

## Explore Numerai

- [Create an account](https://numer.ai/signup): Join the tournament for free.
- [Official onboarding](https://numer.ai/onboarding): Guided tutorials for a first model and submission.
- [Data](https://numer.ai/data): Browse and download available datasets.
- [Leaderboard](https://numer.ai/leaderboard/data-scientists): Current data scientist rankings.
- [Models leaderboard](https://numer.ai/leaderboard/models): Compare live model performance.
- [Latest round](https://numer.ai/round/latest): Current round status and scores.
- [FAQ](https://docs.numer.ai/numerai-tournament/faq): Answers to common participation, data, modeling, staking, and payout questions.

## Official ecosystem

- [NumerAPI](https://github.com/numerai/numerapi): Open-source Python client for data downloads and submissions.
- [Numerai CLI](https://docs.numer.ai/numerai-tournament/submissions/numerai-cli): Open-source tooling for automating cloud model workflows.
- [Numerai Forum](https://forum.numer.ai/): Long-form research and community discussion.
- [Numerai Discord](https://discord.gg/numerai): Official community and support chat.

## Other Numerai competitions

- [Numerai Signals](https://signals.numer.ai/): Bring your own stock-market data and submit original equity signals.
- [Numerai Crypto](https://crypto.numer.ai/): Bring your own crypto data and predict token returns.