DCA Trading Bot

A sophisticated Dollar-Cost Averaging (DCA) bot for cryptocurrency markets with backtesting, parameter optimization, and risk management features. Designed for Binance exchange with testnet support.

Features

• Historical Data Management

  • Automated OHLCV data fetching
  • SQLite storage with deduplication
  • Multiple timeframe support

• Strategy Backtesting

  • Parameter grid search optimization
  • Advanced metrics calculation (Sharpe Ratio, Max Drawdown)
  • Vectorized backtesting engine

• Live Trading

  • Testnet (paper trading) support
  • Real-time order execution
  • Partial fill handling
  • Multi-threaded architecture

• Risk Management

  • Portfolio drawdown monitoring
  • Emergency liquidation triggers
  • Position size optimization

• Adaptive Logic

  • Daily parameter re-optimization
  • Shadow trading validation
  • Strategy health checks

Requirements

• Python 3.8+
• Binance API keys (testnet supported)
• CCXT library
• Pandas/Numpy for quantitative analysis
• SQLite for local storage

Installation

1. Clone repository:

   git clone https://github.com/aaurelions/DCABot
   cd DCABot

2. Install dependencies:

   pip install ccxt pandas numpy sqlite3

3. Create configuration file (config.json):

   {
     "apiKey": "your_api_key",
     "secret": "your_api_secret",
     "testnet": true
   }

4. Initialize database:

   from dca_bot import DCABot
   bot = DCABot()
   bot.insert_test_config()  # Creates sample configuration

Configuration Guide

Token Configuration Schema

Parameter	Description	Default
base_order	Initial position size (USD)	100
dca_order	Follow-up DCA size (USD)	50
price_deviation	Required drop for DCA trigger (%)	2.0
take_profit	Profit target (%)	5.0
max_dca_orders	Maximum DCA levels	5
dca_multiplier	DCA spacing multiplier	1.2
cooldown	Minimum time between rounds (sec)	3600

Database Structure

erDiagram
    tokens_config ||--o{ dca_rounds : symbol
    dca_rounds ||--o{ dca_trades : round_id
    historical_data {
        string symbol
        datetime timestamp
        float open
        float high
        float low
        float close
        float volume
    }
    backtest_results {
        string symbol
        string params_hash
        float total_profit
        float win_rate
        float max_drawdown
        float sharpe_ratio
    }

Loading

Usage

Basic Operation

# Initialize bot
bot = DCABot(config_path='config.json')

# Fetch historical data (1-minute resolution, 100 days)
bot.fetch_historical_data('BTC/USDT', '1m', days=100)

# Run backtest with parameter grid
param_ranges = {
    'price_deviation': [1.5, 2.0, 2.5],
    'take_profit': [3.0, 4.0, 5.0],
    'dca_multiplier': [1.0, 1.2, 1.5]
}
best_params = bot.backtest_strategy('BTC/USDT', param_ranges)

# Start live trading if profitable
bot.execute_dca_strategy_if_profitable('BTC/USDT', param_ranges)

Advanced Features

Daily Optimization Loop

# Runs parameter optimization every 24h
bot.optimize_parameters()

Portfolio Protection

# Force close all positions if 10% drawdown
bot._check_portfolio_drawdown(threshold=0.1)

Shadow Trading

# Validate strategy with fresh market data
bot.run_backtest_validation('ETH/USDT')

Backtesting Methodology

1. Data Preparation

   • Clean OHLCV data with outlier detection
   • Slippage modeling (0.02% default)
   • Exchange fee simulation (0.1% taker fee)

2. Strategy Logic

   graph TD
       A[Base Order] -->|Price < Avg-Avg*Deviation| B[DCA Order]
       A -->|Price > Avg+Avg*TP| C[Take Profit]
       B --> D{Max DCA Reached?}
       D -->|Yes| E[Emergency Close]
       D -->|No| A
   

   Loading

3. Metrics Calculation

   • Sharpe Ratio: Risk-adjusted returns
   • Max Drawdown: Worst peak-to-trough decline
   • Win Rate: Profitable trade percentage

Risk Management

• Position sizing based on volatility
• Circuit breakers for extreme market moves
• Time-based order expiration
• Exchange connectivity monitoring

Disclaimer

This software is for educational purposes only. Cryptocurrency trading involves significant risk of loss and is not suitable for all investors. Past performance is not indicative of future results.

License

MIT License