TSMOM Trading Strategy — Multi-Asset Backtesting

A session-aware Time-Series Momentum (TSMOM) strategy backtested across Gold (XAUUSD), Bitcoin (BTCUSD), and WTI Crude Oil using minute-level price data.

Built as part of the ASIM course at the University of Cologne.

Strategy Design

The strategy trades intraday momentum relative to the previous session's closing price, with three key mechanisms:

• Session-based entries — trades are evaluated independently across Asian, London, and US sessions, each with defined open/close windows
• Multi-attempt scaling — up to 3 entry attempts per session with increasing take-profit targets (1x, 2x, 3x base risk)
• Percentage-based position sizing — 1% capital risk per trade, with stop-loss and take-profit derived from a fixed base risk percentage

Project Structure

File	Purpose
strategy.py	Core TSMOM strategy with session handling, signal generation, and trade simulation
backtest.py	Backtesting engine with per-session metrics, attempt analysis, and fee sensitivity
market_analysis.py	Market regime classification using EMA crossovers and ATR
data_handler.py	Multi-asset data loading and preprocessing pipeline
results/	Pre-computed backtest outputs at 0%, 0.5%, and 1% fee levels

Setup

git clone https://github.com/JuliusScheuerer/ASIM_TSMOM_Trading_Strategy.git
cd ASIM_TSMOM_Trading_Strategy
python -m venv venv && source venv/bin/activate
pip install pandas numpy scipy

Usage

Run the backtest (configure asset and fee levels in backtest.py):

python backtest.py

Analyze market regimes:

python market_analysis.py

Data

The strategy expects minute-level OHLCV data in CSV format under datasets/raw/<ASSET>/. Raw data is excluded from the repository via .gitignore — see data_handler.py for the expected directory structure and file naming conventions.

License

MIT