moss-trade-bot-skills  by moss-site

LLM-powered trading agents from natural language

Created 2 months ago
354 stars

Top 78.6% on SourcePulse

GitHubView on GitHub
Project Summary

Summary

Moss-trade-bot-skills offers an LLM-powered factory to generate, backtest, and evolve cryptocurrency trading agents from natural language. It targets quantitative traders and AI enthusiasts, automating strategy creation and enabling continuous agent self-improvement.

How It Works

The system translates natural language trading styles into five core strategy pillars: Trend, Mean-Reversion, Momentum, Volume, and Risk. It utilizes a robust local backtesting engine with cross-margin simulation and a unique "Weekly Evolution Loop." This loop embeds AI-driven reflection during backtests, analyzing performance to micro-adjust tactical parameters while keeping core personality traits locked. A Signal Decision System normalizes market dimensions into a composite score.

Quick Start & Requirements

Installation requires cloning a specific Git tag (v1.0.28) and installing Python dependencies (pandas>=2.0.0, numpy>=1.24.0, ccxt>=4.0.0). Historical Hyperliquid datasets are automatically downloaded and cached locally. Key commands include fetch_data.py for data fingerprinting, run_backtest.py, and run_evolve_backtest.py for strategy evaluation.

Highlighted Details

  • Natural Language Strategy Generation: Infers parameters, leverage, and indicators from textual descriptions.
  • Advanced Backtesting: Features cross-margin simulation, regime detection, and metrics aligned with live Hyperliquid conditions.
  • Self-Evolving Agents: The "Weekly Evolution Loop" enables continuous adaptation based on performance reflection.
  • Live Copy Trading & Dashboards: Supports live trading for major tokens on Hyperliquid with integrated tracking dashboards.
  • Safety Guardrails: Includes leverage limits, mandatory stop-losses, and confirmation gates.

Maintenance & Community

Actively updated, with the latest release v1.0.28 (June 1, 2026). Recent updates focus on data handling, distribution, and live trading features. No specific community links or contributor information were detailed.

Licensing & Compatibility

Licensed under the permissive MIT-0 License, generally allowing for commercial use and integration into closed-source projects.

Limitations & Caveats

Explicitly for research and educational purposes; trading performance varies and is not financial advice. Live trading and advanced features are optional integrations with the Moss platform.

Health Check
Last Commit

1 week ago

Responsiveness

Inactive

Pull Requests (30d)
0
Issues (30d)
0
Star History
141 stars in the last 30 days

Explore Similar Projects

Feedback? Help us improve.