nstrade

NS Learnathon - Trading for Bitcoin

Setup

Prerequisites

1. Install Homebrew

   • Visit brew.sh

2. Install uv

   # On macOS
   brew install uv

   Note: uv is like poetry but better in every way. Learn more

3. Fork and Clone

   # Fork the repo at https://github.com/athon-millane/nstrade.git
   git clone https://github.com/your-username/nstrade.git

4. Python Setup

   # Optional: Install Python 3.12 if needed
   uv python install 3.12
   
   # Initialize venv and install dependencies
   uv sync

Syncing with Upstream

# Add upstream remote
git remote add upstream https://github.com/athon-millane/nstrade.git

# Pull from upstream (either method)
git pull upstream main

# OR
git fetch upstream
git merge upstream main

Creating a New Strategy

1. Copy the Template

   cp strategies/template.py strategies/my_strategy.py

2. Edit Your Strategy

   • Set your name, strategy name, and description
   • Implement your strategy logic in process_bar() and get_signal()
   • Optionally implement a vectorized version in get_signals()

3. Quick Test

   python scripts/test_strategy.py strategies/my_strategy.py

   This will:

   • Validate your strategy implementation
   • Test on development period (2011-2024)
   • Test on holdout period (2025)

4. Detailed Evaluation

   • Open notebooks/evaluate.ipynb in Jupyter
   • Change the strategy_path to point to your strategy file
   • Run all cells to see:
     • Detailed performance metrics
     • Interactive plots of price, volume, equity curve, drawdowns, and Sharpe ratio
     • Buy/sell points overlaid on the price chart
     • Trade-by-trade analysis

5. Submit Your Strategy

   • Create a pull request
   • Our CI will validate and backtest your strategy
   • If successful, it will be added to the leaderboard

Project Status

Completed

• Speed up backtest (maybe vectorise, maybe parallelise)
• Set up a standardised flow for creating, contributing and backtesting new Strategy
• Capture user metadata in Strategy for leaderboard display

To Do

• Set up CI/CD YAML file:
  • Validate strategy backtestability
  • Run backtests on our dataset
  • Store results in leaderboard file
  • Handle unique submissions (User + Strategy name)
  • Define development/holdout sets:
    • development.csv: Data up to end of 2024
    • holdout.csv: All of 2025
    • Strategies should only use development.csv

Metrics

Current Focus

• Sharpe ratio as primary metric
• Drawdown
• Win rate
• Number of trades

Future Metrics

• Sortino ratio
• R² of equity curve (suggested by Barnabas)

Notes

Security

• Current focus: Ensure test data isn't used in strategy development
• More comprehensive security measures to be addressed later

Observations

• SMA Crossover performance appears suspiciously good
• Most bootstrapped strategies perform poorly without fees and slippage
• Recommended: Include 0.3% fee+slippage per trade