🌙 nof1-agents - Autonomous Crypto Trading Bot

AI-powered autonomous trading inspired by Nof1.ai Alpha Arena
Pure LLM decision-making + HyperLiquid perpetuals = Fully autonomous crypto trading

This project is a clone/implementation inspired by Nof1.ai's Alpha Arena approach, where AI models make autonomous trading decisions based purely on technical analysis and market data.

---

🎯 What is This?

nof1-agents is an autonomous cryptocurrency trading system that uses Large Language Models (LLMs) to:

• Analyze real-time market data from HyperLiquid perpetuals exchange
• Make independent trading decisions (LONG/SHORT/CLOSE/HOLD)
• Execute trades automatically with risk management
• Track performance and log all reasoning

How It Works

Every 5 minutes (configurable), the AI:

1. 📊 Collects Data - Account balance, positions, market data (OHLCV, indicators, funding rates)
2. 🤖 Analyzes - Sends data to DeepSeek LLM with technical context
3. 🎯 Decides - AI responds with trading decision + confidence + reasoning
4. ⚡ Executes - If confidence ≥ 65%, executes trade on HyperLiquid
5. 📝 Logs - Saves reasoning traces, trades, and performance

Example AI Decision:

{
  "decision": "OPEN_LONG",
  "symbol": "BTC",
  "confidence": 0.88,
  "reasoning": "Strong uptrend, RSI recovering from oversold, MACD bullish crossover...",
  "position_size_usd": 9000,
  "leverage": 20,
  "stop_loss": 105000,
  "take_profit": 115000
}

---

⭐ Key Features

• 🤖 Pure LLM Trading - No hard-coded rules, AI decides everything
• 📈 Multi-Asset - Trade BTC, ETH, SOL, BNB, DOGE, XRP
• ⚡ High Leverage - Up to 50x leverage on HyperLiquid (configurable)
• 🛡️ Risk Management - Stop loss, take profit, position sizing, confidence thresholds
• 📊 Technical Analysis - EMA, RSI, MACD, ATR, Bollinger Bands, Funding Rates
• 🔄 Modular Design - Swap AI models easily (DeepSeek, Claude, GPT-4, Gemini)
• 📝 Complete Logging - All reasoning, trades, and performance tracked
• 🎨 Nof1-Style Prompts - Inspired by Alpha Arena's technical approach

---

🚀 Quick Start

Prerequisites

Python 3.10 Required (due to dependency compatibility)

Check your Python version:

python3.10 --version

If you don't have Python 3.10, install it:

macOS (Homebrew):

brew install python@3.10

Ubuntu/Debian:

sudo apt update
sudo apt install python3.10 python3.10-venv python3.10-dev

Windows:
Download from python.org

---

1. Clone & Setup

# Clone the repository
git clone <your-repo-url>
cd nof1-agents

# Create virtual environment with Python 3.10
python3.10 -m venv venv

# Activate virtual environment
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

---

2. Configure API Keys

# Copy environment template
cp env.example .env

# Edit .env with your actual keys
nano .env  # or use your preferred editor

Required API Keys in .env:

# Get DeepSeek API key from: https://platform.deepseek.com
DEEPSEEK_KEY=your_deepseek_api_key_here

# Your HyperLiquid Ethereum private key (NEVER SHARE THIS!)
HYPER_LIQUID_ETH_PRIVATE_KEY=0xyour_private_key_here

# Your HyperLiquid wallet address
HYPER_LIQUID_MASTER_ADDRESS=0xyour_wallet_address_here

Where to Get Keys:

• DeepSeek API: Sign up at platform.deepseek.com
• HyperLiquid: Use your existing Ethereum wallet's private key and address

⚠️ SECURITY WARNING: Never commit your .env file to Git! It's already in .gitignore.

---

3. Configure Trading Settings

Edit agents/agent_configs.py to customize your trading parameters:

CONFIG = {
    'STARTING_CAPITAL_USD': 500,         # Your starting capital
    'SYMBOLS': ['BTC', 'ETH', 'SOL'],    # Coins to trade
    'MAX_LEVERAGE': 20,                   # Maximum leverage (1-50x)
    'MAX_POSITION_PERCENT': 90,          # Use 90% of capital per position
    'CHECK_INTERVAL_MINUTES': 3,          # How often to check (minutes)
    'MIN_CONFIDENCE_TO_TRADE': 0.65,      # Only trade if AI ≥65% confident
    'STOP_LOSS_PERCENT': 5.0,             # Auto-exit at -5% loss
    'TAKE_PROFIT_PERCENT': 10.0,          # Auto-exit at +10% gain
    'MODEL_NAME': 'deepseek-chat',        # AI model to use
    'TEMPERATURE': 0.7,                   # LLM creativity (0-1)
}

💡 Recommended Starting Settings:

# For beginners - start with these safer settings:
CONFIG = {
    'STARTING_CAPITAL_USD': 100,         # Start small
    'SYMBOLS': ['BTC', 'ETH'],           # Just 2 coins
    'MAX_LEVERAGE': 5,                    # Low leverage!
    'CHECK_INTERVAL_MINUTES': 5,          # Less frequent
    'MIN_CONFIDENCE_TO_TRADE': 0.75,      # Higher confidence
}

🎯 Quick Presets Available:

You can use pre-configured presets in agent_configs.py:

• CONSERVATIVE_CONFIG - Lower risk

  • 10x leverage (vs 20x default)
  • 50% position size (vs 90%)
  • 75% confidence required (vs 65%)
  • 3% stop loss (tighter)

• AGGRESSIVE_CONFIG - Higher risk

  • 30x leverage (vs 20x default)
  • 95% position size (vs 90%)
  • 60% confidence required (vs 65%)
  • 7% stop loss (wider)

• MULTI_SYMBOL_CONFIG - Diversified

  • Trade 3 symbols simultaneously
  • 30% per symbol (90% total)
  • 5 minute intervals

---

4. Run the Trading Bot

# Make sure virtual environment is activated
source venv/bin/activate

# Run the trader
python agents/deepseek_trader.py

# Or use the standalone runner
python run.py

You should see:

🌙 DeepSeek Nof1 Trader - Alpha Arena Style
✅ DeepSeek ready: deepseek-chat
📋 DeepSeek Trader Configuration loaded
🚀 Starting Nof1 Trading Agent...

Stop the bot: Press Ctrl+C

---

📁 Project Structure

nof1-agents/
├── agents/                     # 🤖 Trading agent logic
│   ├── deepseek_trader.py      # Main trading bot (RUN THIS)
│   ├── base_pure_agent.py      # Base agent class
│   └── agent_configs.py        # ⚙️ Configuration (EDIT THIS)
│
├── models/                     # 🧠 LLM implementations
│   ├── model_factory.py        # Model selector
│   ├── deepseek_model.py       # DeepSeek integration
│   ├── claude_model.py         # Claude integration
│   ├── openai_model.py         # GPT integration
│   └── ...                     # Other models
│
├── exchange/                   # 💱 Exchange integrations
│   └── nice_funcs_hyperliquid.py  # HyperLiquid API wrapper
│
├── data_formatters/            # 📊 Data formatting
│   ├── market_data_formatter.py   # Market data + indicators
│   ├── position_formatter.py      # Position tracking
│   └── account_formatter.py       # Account metrics
│
├── prompts/                    # 💬 LLM prompts
│   └── nof1_style_prompt.py    # Nof1-inspired prompt format
│
├── requirements.txt            # 📦 Dependencies
├── env.example                 # 🔑 Environment template
├── run.py                      # 🚀 Standalone runner
└── README.md                   # 📖 This file

---

⚙️ Configuration Options

All settings in agents/agent_configs.py:

| Setting | Default | Description | |---------|---------|-------------| | STARTING_CAPITAL_USD | 500 | Your starting capital ($) | | SYMBOLS | BTC, ETH, SOL, BNB, DOGE, XRP | Coins to trade | | MAX_LEVERAGE | 20 | Maximum leverage (1-50x) | | MAX_POSITION_PERCENT | 90 | % of capital per position | | CHECK_INTERVAL_MINUTES | 3 | Check every N minutes | | MIN_CONFIDENCE_TO_TRADE | 0.65 | Only trade when AI ≥65% confident | | STOP_LOSS_PERCENT | 5.0 | Auto-exit at -5% loss | | TAKE_PROFIT_PERCENT | 10.0 | Auto-exit at +10% gain | | MODEL_NAME | deepseek-chat | LLM model to use | | TEMPERATURE | 0.7 | LLM creativity (0-1) |

Example: Conservative Trading

from agents.deepseek_trader import DeepSeekTrader

trader = DeepSeekTrader(
    symbols=['BTC', 'ETH'],
    starting_capital_usd=100,
    max_leverage=10,           # Lower leverage
    min_confidence=0.75,       # Higher confidence required
    stop_loss_percent=3.0,     # Tighter stop loss
    check_interval_minutes=5   # Check less frequently
)

trader.run()

---

📊 Logs & Performance Tracking

All data saved to src/data/nof1_agents/:

📝 Reasoning Traces

src/data/nof1_agents/reasoning/deepseek_2025-10-27.txt

Contains full AI reasoning for every decision:

[2025-10-27 14:32:15] DEEPSEEK REASONING TRACE
DECISION: OPEN_LONG BTC
CONFIDENCE: 88%

REASONING:
Strong bullish momentum observed:
1. Price broke above 20-EMA with conviction
2. RSI at 45, recovering from oversold
3. MACD showing bullish crossover
4. Funding rate slightly negative (long opportunity)
5. Volume increasing on upward moves

TRADE PARAMETERS:
- Entry: $111,113.50
- Position: $9,000 (20x leverage)
- Stop Loss: $105,000
- Take Profit: $115,000
- Risk/Reward: 3.5

💰 Trade Logs

src/data/nof1_agents/trades/2025-10-27_trades.json

JSON log of all executed trades.

📈 Performance Metrics

src/data/nof1_agents/performance/

P&L tracking, Sharpe ratio, win rate, drawdown.

---

🛡️ Risk Management

Built-in Protections

• ✅ Confidence Thresholds - Only trade when AI is confident enough
• ✅ Position Sizing - Limits per-trade exposure
• ✅ Stop Loss - Automatic exit on adverse moves
• ✅ Take Profit - Lock in gains automatically
• ✅ Leverage Caps - Prevent over-leveraging
• ✅ Max Drawdown - Stop trading if down X%
• ✅ Min Balance - Preserve minimum account balance

⚠️ Important Warnings

• ⚠️ You can lose all your capital - High leverage = high risk
• ⚠️ AI can make mistakes - No system is perfect
• ⚠️ Start small - Test with $10-50 first
• ⚠️ Monitor closely - Watch the first few hours