Updated:
Numerai
Richard Craib's Numerai runs a crowd-sourced hedge fund where data scientists stake its NMR token on encrypted market predictions.
Numerai
Numerai was established in 2015 by Richard Craib, a South African mathematician who left a quantitative trading role to build what he called 'the world's first crowd-sourced hedge fund.' The firm does not disclose its underlying wealth origin or backing, operating instead as a standalone asset manager that has raised capital from venture investors including Union Square Ventures and Paradigm to fund its development and initial fund seeding. Numerai runs a fundamentally unorthodox investment strategy. It obfuscates its proprietary market data, encrypts it, and releases it as abstracted datasets to thousands of anonymous data scientists around the world who compete in weekly tournaments to build the most predictive machine-learning models. The firm aggregates these models into a meta-model that drives its trading across global equities. This is not a fund-of-funds or a traditional quant fund—it is a hybrid that pays modelers in its native cryptocurrency, NMR, which they must stake on their predictions, aligning incentives without revealing source data. Known structural participants include the firm itself operating its principal fund and the stakers who bond NMR to signal confidence in specific models. Team size and total assets under management remain undisclosed; the firm's operational scale is instead publicly visible through its smart contract activity. Staked capital has exceeded $50 million, with the NMR token market capitalization fluctuating with market conditions. In late 2023, Numerai launched Numerai Signals, a separate tournament that allows participants to submit predictions based on their own external data, expanding the model universe beyond the original encrypted dataset tournament. The firm operates primarily from San Francisco. Numerai's structural differentiator is its cryptographic core: it is a hedge fund that cannot see the strategies of its contributing data scientists, and its data scientists cannot see the assets they are predicting. This dual-blind architecture—enforced by homomorphic encryption and a staking token—means the fund's edge is an ever-evolving ensemble of models from an uncoordinated global network. No other quant fund operates with this combination of zero-knowledge proof-style data science and tokenized staking on blockchain rails.
General information
Firm type
Asset Manager
Year founded
2015
AUM
Undisclosed
Location
Region
North America
Country
United States
City
San Francisco
Corporate office
San Francisco, CA, United States
Principals
Richard Craib
Founder
Sector focus
Frequently asked questions
How does Numerai source its investment models?
Numerai runs a weekly tournament in which it releases encrypted, obfuscated market data to a global community of data scientists. Participants build machine-learning models on this data—without knowing what the underlying features represent—and submit predictions. Numerai aggregates the top-performing models into a meta-model that drives its proprietary trading. In 2023, it expanded this model with Numerai Signals, allowing submissions based on participants' own external datasets.
What is the role of the NMR token in Numerai's structure?
NMR is an Ethereum-based token that serves as a work and staking asset within the Numerai ecosystem. Data scientists earn NMR for submitting successful predictions, but they must also stake NMR on their models to signal confidence—if their models perform poorly, staked NMR is burned. The token aligns incentives by creating a direct financial linkage between a model's marginal contribution and its creator's skin in the game, while enabling a decentralized, anonymous workforce.
Who are Numerai's known backers or strategic investors?
Numerai has raised capital from venture investors including Union Square Ventures and Paradigm, a crypto-focused investment firm. These investors are not participants in the tournament or the hedge fund's profits but early financial backers of the company Numerai Inc., which operates the infrastructure, develops the cryptography, and manages the principal fund.
How does Numerai protect its proprietary data from the data scientists who build its models?
Numerai employs a form of homomorphic encryption: the firm transforms raw market data into an abstracted, encrypted dataset of features that statistical relationships can be discovered within, but which cannot be reverse-engineered to reveal the original assets, prices, or time periods. This allows the firm to pay outside modelers while ensuring that even the most successful contributors cannot replicate the strategy elsewhere or identify the specific trades the fund makes.
Is Numerai structured as a decentralized autonomous organization (DAO)?
No. Numerai Inc., the San Francisco-based company founded by Richard Craib, operates the tournament, manages the meta-model, executes trades, and controls the master fund. The staking and payment mechanisms are decentralized on Ethereum, but the fund itself is a centralized quant manager that uses the decentralized network as an input—not a DAO-governed investment vehicle.
What asset classes does Numerai trade?
Numerai focuses on global equities. The firm has stated that the original tournament dataset is derived from equity market data, and its principal fund trades long-short equity positions across global markets. The firm has not publicly disclosed any expansion into fixed income, commodities, digital assets, or private markets for its principal fund.
What differentiates Numerai Signals from the original Numerai tournament?
Numerai Signals, launched in late 2023, lets data scientists submit predictions trained on their own independently acquired datasets—alternative data, public market feeds, or proprietary research—rather than on Numerai's encrypted dataset. It operates as a parallel tournament that feeds into the same meta-model, allowing the fund to capture predictive signals from data sources it does not own or curate.
Profile maintained by Altss using OSINT (open-source intelligence), regulatory filings, licensed data partners, and verified direct submissions. Read the methodology. Last updated: . Continuous refresh with full update cycles at least every 30 days.
Need institutional-grade insight on family offices?
Altss delivers:
Prefer a guided tour?
We’ll walk you through: