ADRPairsTrading

Report

• report.pdf is the final report of our stock-ADR pairs trading strategy.

Data Pulling and Engineering

• historical_data contains the stock and forex data used in backtesting.
• historical_data/pull_data.ipynb is used to select our universe of stock-ADR pairs, and to extract historical stock and forex data of interest.
• historical_data/filter_forex.ipynb is used to filter the forex data pulled, for backtesting.
• market_metadata contains the interest rate and vix data used to select a suitable outsample period.
• market_metadata/pull_vix.ipynb and market_metadata/generate_interest_rates.ipynb are used to obtain interest rate and vix data over our backtesting period.

Backtesting

• strategies.py contains the code to run all 4 variants of our pairs trading strategy, with local risk measures and realistic market assumptions, on a single pair.
• helpers.py contains code that estimates the ADR:ORD ratio, merges raw data into a single dataframe, and calculates useful statistics.
• market_metadata/select_outsample.ipynb is used to select our outsample period used in backtesting.
• local_pairs_hp_tune.ipynb conducts hyperparameter tuning to determine the best hyperparameters for each adr-stock pair, for each variant, on our in-sample period.
• final_strategy.ipynb selects the optimal variant, selects the pairs we wish to trade in our portfolio, defines the strategy that trades multiple pairs at once, incorporates global risk measures for the entire portfolio, conducts hyperparameter tuning for portfolio allocation, and reports final results.

Live Trading

• live_trading_implementation.ipynb updates data periodically, decides what trades to make and is used to place trades.

Visualizations

• Visualizations.ipynb generates all the graphs that we use in our final report, which are stored in the visualizations folder.