An open-source platform for evaluating AI cryptocurrency trading models in live markets with real capital. It targets researchers and developers seeking to benchmark AI performance beyond static datasets, offering a transparent, real-time trading environment. The project democratizes AI trading research and serves as an educational tool.

How It Works

The platform automates cryptocurrency trading on Binance using the CCXT library. AI models, primarily DeepSeek, analyze market data and execute trades every three minutes after a chain-of-thought reasoning process. Account metrics are collected every twenty seconds. This approach contrasts with traditional benchmarks by testing AI capabilities in dynamic, adversarial, real-world financial conditions, emphasizing risk management and strategic decision-making.

Quick Start & Requirements

• Installation: Clone the repository, run bun install.
• Prerequisites: Bun, PostgreSQL, Binance API credentials, DeepSeek API key. Optional: OpenRouter API key, Exa API key.
• Setup: Configure .env with API keys and database URL. Generate and push the database schema using bunx prisma generate and bunx prisma db push.
• Running: Start the development server with bun dev.
• Automation: Requires external cron job setup for 20-second metric collection and 3-minute trading execution endpoints.
• Links: Repository: https://github.com/snowingfox/open-nof1.ai.git.

Highlighted Details

• Real-time trading execution on Binance via CCXT.
• Live dashboard featuring account performance charts and detailed trade history.
• AI decision-making powered by DeepSeek models with transparent chain-of-thought reasoning.
• Supports trading for BTC, ETH, SOL, BNB, and DOGE.
• Tech stack includes Next.js 15, Bun, Prisma, Tailwind CSS v4, and Vercel AI SDK.

Maintenance & Community

The README does not specify maintainers, community channels (like Discord or Slack), or active sponsorships. It is presented as an open-source implementation inspired by the closed nof1.ai Alpha Arena.

Licensing & Compatibility

The project is released under the MIT License, which permits broad use, modification, and distribution, including for commercial purposes and integration into closed-source applications.

Limitations & Caveats

The "Positions" feature for tracking open trades is marked as "coming soon." The software is intended for educational and research purposes, and trading cryptocurrencies carries substantial risk of loss. AI model decisions may be suboptimal, and past performance is not indicative of future results. Automated trading requires manual configuration of external cron jobs.