Numerai ends a decade of stock anonymity with Unity dataset, merging tournaments
matlabulous · x · 2026-10-17
Quant hedge fund Numerai announced a historic shift: after ten years of hiding stock identities, it launched the Unity dataset with real stock identifiers and is merging Numerai and Numerai Signals into one tournament.
- Unity dataset: 1,040 features plus real stock identifiers, doubling the NCORR of the Numerai Signals benchmark over validation; features stay ranked, normalized, binned and unnamed — only the rows now map to real companies.
- Tournament merge: both run separately for a few months, then unify on January 1, 2027 into one tournament with one Meta Model, using Jupiter targets and current neutralizers.
- Built for agents: 77,000 MCP sessions since July and 30,000+ in the last 30 days. Until now agents could only do half the job (modeling); Unity lets them work the whole pipeline — sourcing and engineering features included.
It marks crowdsourced quant moving from humans modeling anonymized data to fleets of autonomous AI agents researching global equities.
More from Research
- visloc-rs: pure-Rust SfM 3.46x faster than COLMAP, beats ORB-SLAM3 on 9/11 EuRoC — rsasaki0109 · 2026-10-17
- AgentForesight: a 7B online auditor that catches multi-agent failures mid-run — jiqizhixin · 2026-10-17
- Researchers stranded an AI agent on a virtual island — and things got weird — UFOsAreAGIs · 2026-10-17
- Yann LeCun and NYU team propose AdaJEPA to adapt world models when the world changes — mengyer · 2026-10-17
- Harvard mathematician open-sources attempt to get Claude to construct Yang-Mills — lvwerra · 2026-10-17
- Model tweaks barely move MMLU but easily break generative tasks — aryaman2020 · 2026-10-17