🤖 AI Trading Bot

A modular, multi-agent AI trading bot for BTC/USDT on Binance with pluggable ML/AI support.

---

⚡ Architecture

                    ┌──────────────┐
                    │  run_bot.py  │
                    └──────┬───────┘
                           │
                    ┌──────▼───────┐
                    │ Orchestrator │ ← Main async loop
                    └──┬──┬──┬──┬─┘
          ┌────────────┘  │  │  └────────────┐
          ▼               ▼  ▼               ▼
  ┌───────────────┐ ┌─────────────┐ ┌──────────────┐
  │ Market Data   │ │  Feature    │ │  Strategy    │
  │ Agent         │ │  Agent      │ │  Agent       │
  └───────┬───────┘ └──────┬──────┘ └──────┬───────┘
          │                │               │
          ▼                ▼               ▼
  ┌───────────────┐ ┌─────────────┐ ┌──────────────┐
  │ Binance API   │ │ RSI/EMA/    │ │ Trend/MeanR/ │
  │ + SQLite      │ │ MACD        │ │ Momentum+ML  │
  └───────────────┘ └─────────────┘ └──────────────┘
          │                               │
          │         ┌─────────────┐       │
          │         │  Risk Agent │◄──────┘
          │         └──────┬──────┘
          │                │
          │         ┌──────▼──────┐
          └────────►│ Execution   │──► Binance Orders
                    │ Agent       │
                    └─────────────┘

🚀 Quick Start

1. Clone & Install

cd Ai-trading-bot
pip install -r requirements.txt

2. Configure

cp .env.example .env
# Edit .env with your Binance API keys

3. Run

# Paper trading (testnet — no real money)
python scripts/run_bot.py --paper

# Live trading (requires API keys + confirmation)
python scripts/run_bot.py --live

# Custom symbol / interval
python scripts/run_bot.py --paper --symbol ETHUSDT --interval 15m

---

📂 Project Structure

ai-trading-bot/
├── agents/                    # Trading agents
│   ├── market_data_agent.py   # Fetches OHLCV from Binance
│   ├── feature_agent.py       # Computes RSI, EMA, MACD
│   ├── strategy_agent.py      # Runs strategies, aggregates signals
│   ├── risk_agent.py          # Position sizing, stop-loss, limits
│   └── execution_agent.py     # Places orders on Binance
│
├── strategies/                # Trading strategies
│   ├── trend_strategy.py      # Dual EMA crossover
│   ├── mean_reversion.py      # Bollinger Bands
│   └── momentum_strategy.py   # RSI + MACD dual confirmation
│
├── indicators/                # Technical indicators
│   ├── rsi.py                 # Relative Strength Index
│   ├── ema.py                 # Exponential Moving Average
│   └── macd.py                # MACD line/signal/histogram
│
├── exchange/                  # Exchange integration
│   └── binance_client.py      # python-binance wrapper
│
├── ml/                        # AI/ML integration
│   ├── base_model.py          # Abstract model interface
│   └── sklearn_model.py       # RandomForest example
│
├── core/                      # Core components
│   ├── config.py              # Environment-based config
│   └── orchestrator.py        # Main trading loop
│
├── data/                      # Auto-created data dir
│   └── market_cache.db        # SQLite kline cache
│
├── scripts/
│   └── run_bot.py             # CLI entry point
│
├── .env.example               # Config template
├── requirements.txt           # Python dependencies
└── README.md

---

🧠 Strategies

Strategy	Method	Buy Signal	Sell Signal
Trend	EMA Crossover (9/21)	Fast EMA > Slow EMA	Fast EMA < Slow EMA
Mean Reversion	Bollinger Bands (20, 2σ)	Price ≤ Lower Band	Price ≥ Upper Band
Momentum	RSI + MACD	RSI < 30 + MACD bullish cross	RSI > 70 + MACD bearish cross

Signals include a confidence score (0–1). When multiple strategies agree, confidence is boosted.

---

🛡️ Risk Management

• Position sizing — Risk only RISK_PER_TRADE % of balance per trade
• Max position — Never exceed MAX_POSITION_PCT % of balance
• Stop-loss / Take-profit — Automatically calculated per trade
• Daily loss limit — Bot stops trading if daily loss exceeds MAX_DAILY_LOSS %
• Trade cooldown — Minimum TRADE_COOLDOWN_SECONDS between trades
• Confidence threshold — Signals below 30% confidence are rejected

---

🤖 AI/ML Integration

The bot supports pluggable ML models. To add your own:

1. Subclass ml.base_model.BasePredictionModel
2. Implement train(), predict(), save(), load()
3. Train your model and save it
4. Set USE_ML_MODEL=true and ML_MODEL_PATH=path/to/model.pkl in .env

Supported Frameworks

• ✅ scikit-learn (included example)
• ✅ TensorFlow / Keras
• ✅ PyTorch
• ✅ XGBoost / LightGBM
• ✅ Any framework that fits the interface

---

⚠️ Disclaimer

This bot trades with real money in live mode. Cryptocurrency trading involves substantial risk of loss. The developers are not responsible for any financial losses incurred through the use of this software. Always:

• Test thoroughly in paper mode first
• Never risk more than you can afford to lose
• Monitor the bot while it's running
• Start with minimum position sizes