# CrowdCent

> CrowdCent is building the next generation of investing: decentralized, systematized, and democratized. We combine crowdsourced investment insights from fundamental analysts with machine learning models from data scientists to create meta-models that guide real capital allocation. Our mission is to decentralize investment management by harnessing collective intelligence—human expertise amplified by AI.

## Core Products

CrowdCent operates three main products:

1. **CrowdCent Challenge**: Open data science competitions where ML engineers build predictive models on our curated datasets. Models are aggregated into meta-models that power the on-site Simulator, a long/short strategy lab over real Hyperliquid market data with parameter sweeps, multi-sleeve blending, and out-of-sample holdouts, plus live Hyperliquid rebalancing through CrowdCent for deployed strategies (staff preview).

2. **SumZero Capital LP**: A crowdsourced hedge fund that uses NLP/LLMs to extract structured data from fundamental analyst investment pitches, trains ML models on historical submissions, and allocates tens of millions of dollars based on collective intelligence.

3. **CrowdCent NMR LP**: Active participant in the Numerai tournament, staking NMR on multiple models in both the classic tournament and Numerai Signals.

## Philosophy

- **Value Investing + ML**: We pair machine learning with fundamental value principles for long-term horizons, teaching algorithms to recognize business quality that endures
- **Human + Machine Intelligence**: Crowdsourced human insights from investment communities, amplified by crowdsourced ML models from data scientists
- **Quality over Quantity**: We seek model diversity and novel approaches, not volume. A well-designed ensemble of complementary models outperforms thousands of correlated ones.
- **Pure Meritocracy**: Ideas and models judged solely on predictive performance, not pedigree or credentials
- **Collaboration over Competition**: We have no competitors, only potential collaborators

## Getting Started

- [CrowdCent Challenge Documentation](https://docs.crowdcent.com): Complete docs for participating in data science competitions
- [Quick Start Tutorial (20 seconds)](https://docs.crowdcent.com/tutorials/hyperliquid-end-to-end/): Go from zero to your first submission
- [Python API Client](https://docs.crowdcent.com/install-quickstart/): Install and use the crowdcent-challenge client
- [API Reference](https://docs.crowdcent.com/api-reference/python/): Full API documentation

## Active Challenges

- [Hyperliquid Ranking Challenge](https://docs.crowdcent.com/hyperliquid-ranking/): Predict future relative price movement of crypto assets on Hyperliquid at 10-day and 30-day horizons
- [Hyperliquid Ranking Leaderboard](https://crowdcent.com/leaderboard/): Current rankings and performance metrics
- [Meta-Model Simulator](https://crowdcent.com/challenge/hyperliquid-ranking/meta-model/simulation/): Backtest and blend meta-model strategies on real Hyperliquid data; higher CC-Point tiers unlock real-time data, HRP weighting, and risk/alpha controls
- [Scoring System](https://docs.crowdcent.com/scoring/): How submissions are evaluated and ranked

## Open Source Tools

- [NumerBlox](https://crowdcent.github.io/numerblox/): Solid Numerai pipelines - Python library for building robust ML workflows for the Numerai tournament (116 GitHub stars)
- [Centimators](https://crowdcent.github.io/centimators/): Dataframe-agnostic, sklearn-style transformers and ML models for data science competitions (18 GitHub stars)
- [cc-liquid](https://github.com/crowdcent/cc-liquid): Open-source CLI for self-custody Hyperliquid rebalancing on the meta-model; strategy research now lives in the on-site Simulator (8 GitHub stars)
- [crowdcent-challenge Python Client](https://github.com/crowdcent/crowdcent-challenge): Python API client, CLI, and docs for the CrowdCent Challenge (4 GitHub stars, MIT licensed)
- [CrowdCent MCP Server](https://docs.crowdcent.com/ai-agents-mcp/): Built into the crowdcent-challenge package; AI assistants can download data, submit predictions, and backtest strategies on the meta-model. Connect the hosted server at https://mcp.crowdcent.com/mcp with your API key, or run locally via `uvx --from 'crowdcent-challenge[mcp]' crowdcent-mcp`

## About & Mission

- [Our Mission](https://crowdcent.com/mission/): Detailed explanation of how we're decentralizing investment management through collective intelligence
- [About CrowdCent](https://crowdcent.com/about/): Team, culture, values, and inspiration
- [Company Culture](https://crowdcent.com/about/#our-culture): Integrity, First Principles, Accountability, Modesty, Collaboration, Elegant Pragmatism

## Insights & Research

- [Introducing the CrowdCent Challenge](https://crowdcent.com/insights/introducing-the-crowdcent-challenge-the-future-of-investment-management/): Announcement and vision for the future of investment management (June 2025)
- [Machine Learning, Meet Value Investing](https://crowdcent.com/insights/machine-learning-meet-value-investing/): How we pair ML with long-term value investing principles (Feb 2021)
- [CrowdCent <3 Numerai](https://crowdcent.com/insights/crowdcent-3-numerai/): Why we love Numerai and cryptoeconomic incentives (May 2021)
- [Fun with Language Models](https://crowdcent.com/insights/fun-with-language-models/): How we use NLP to enable investment decision-making pipelines (July 2020)
- [Machine Learning for SumZero Data (Whitepaper)](https://storage.googleapis.com/crowdcent-challenge-static/machine_learning_for_sumzero_data.pdf): Technical whitepaper on our approach to alpha capture with crowdsourced fundamental research

## Community

- [Discord Server](https://discord.gg/v6ZSGuTbQS): Join our community for support, discussions, and collaboration
- [Twitter/X](https://twitter.com/crowdcent): Follow us @CrowdCent for updates
- [GitHub Organization](https://github.com/crowdcent): All our open source projects
- [CrowdCent Curation (Substack)](https://curation.crowdcent.com): Curated content across Finance, Blockchain, and Technology
- [Contact Email](mailto:info@crowdcent.com): info@crowdcent.com

## Team

- **Jason Rosenfeld** (Co-Founder & CEO): Full-stack ML engineer and value investor. Former Data Scientist at Millennium Management. Scikit-learn Expert and Numerai Grandmaster. MS in Information and Data Science (UC Berkeley, NLP focus).
- **Ryan Mahon** (Co-Founder): Investor and technologist. Former Portfolio Manager at Millennium Management. CFA Charterholder.

## Optional

- [FAQ](https://docs.crowdcent.com/faq/): Frequently asked questions about the challenge, API keys, and more
- [Contributing Guidelines](https://docs.crowdcent.com/contributing/): How to contribute to our open source projects
- [Disclaimer](https://docs.crowdcent.com/disclaimer/): Legal disclaimer for the challenge
- [Terms of Service](https://crowdcent.com/terms/): Platform terms of service
- [Privacy Policy](https://crowdcent.com/privacy/): How we handle user data
- [Legal](https://crowdcent.com/legal/): Legal information
- [Changelog](https://docs.crowdcent.com/changelog/): Recent updates and changes to the platform
- [LinkedIn](https://www.linkedin.com/company/crowdcent): Company LinkedIn profile
- [Numerai Profile](https://numerai.com/~crowdcent): Our Numerai tournament profile